JP4778076B2 - Component recognition device and component transfer device - Google Patents

Description

  The present invention relates to a component recognition device and a component transfer device, and more particularly to a component recognition device and a component transfer device that perform image recognition based on a captured image of a captured component.

  2. Description of the Related Art Conventionally, a component recognition device and a component transfer device that perform image recognition based on a captured image of a captured component are known (see, for example, Patent Documents 1 and 2).

  In the image processing apparatus (part recognition apparatus) of Patent Document 1 described above, a circumscribed rectangle of a work included in a captured image by a CPU (central processing unit) is obtained based on a captured image of the work (part) captured by the imaging apparatus. Configured to detect. In this image processing apparatus, the contour of the work image included in the captured image is detected, and the envelope point of the work image corresponding to each vertex of the smallest convex polygon including the work image is detected along the contour. Is detected. Specifically, first, the uppermost end point (first envelope point) on the contour line of the work image is determined, and a straight line passing through the envelope point is drawn. The straight line is rotated counterclockwise around the envelope point until it contacts the contour line of the workpiece image, and the contact point between the rotated straight line and the contour line is set as the second envelope point. Thereafter, the same process is repeated until the first envelope point is detected again after making a round of the work image. Thereby, an envelope point of the work image corresponding to each vertex of the polygon including the work image is detected. Based on the detected coordinates of the envelope point, a circumscribed rectangle circumscribing the workpiece at the envelope point is detected.

  Further, the electronic component position detection device (component recognition device) of Patent Document 2 includes an imaging device and a CPU (central processing unit), and is a circumscribed rectangle of a component image detected based on a captured image of the captured component. And the range of the search window for recognizing the position of the feature point of the part in the picked-up image based on the feature point circumscribed rectangle that is the smallest rectangle including all the feature points of the part to be imaged It is configured. The feature point of a component is information on the arrangement and shape of a component terminal (lead, ball, etc.), sides and corners, and is registered in advance. As a result, the range of the feature points in the captured image of the actual part that is imaged in a state including positional deviation and inclination is narrowed down using the arrangement and shape data of the feature points of the parts set and registered in advance. Thus, the feature points of the parts in the captured image are included in the search window.

JP 2007-48194 A JP 2007-58436 A

  However, in the image processing apparatus of Patent Literature 1, as described above, an envelope point is detected until the workpiece image goes around along the contour of the workpiece image included in the captured image, and based on the coordinates of the detected envelope point, Since it is configured to detect a circumscribed rectangle circumscribing the workpiece at the envelope point, the detected workpiece (part) image needs to be a closed region surrounded by a contour line. Therefore, for example, when the captured image includes regions that are the same component but are separated from each other and have a closed outline, such as an image of a component terminal, each terminal is recognized as a separate workpiece. There is an inconvenience. For this reason, there is a problem that a component image composed of a plurality of image portions cannot be recognized as one component.

  In addition, as described above, the electronic component position detection device of Patent Document 2 uses the feature point data of components set and registered in advance, so that an image of a component having a separated image portion is recognized as one component. On the other hand, there is a disadvantage that a large number of feature points including each image portion must be set and registered in advance. In addition, there are many parts that have complicated shapes and it is difficult to set feature points. When handling many types of parts, each part must be set individually, which complicates feature point setting work. There is a problem.

  The present invention has been made in order to solve the above-described problems, and one object of the present invention is to use a component image composed of a plurality of image portions as one component without requiring a complicated setting operation. To provide a component recognition device and a component transfer device that can be recognized.

Means for Solving the Problems and Effects of the Invention

To achieve the above object, a component recognition apparatus according to a first aspect of the present invention performs image recognition based on a captured image including a plurality of image portions corresponding to a component obtained by imaging the component. the provided image recognition unit, in comparison with the dimension data of a predetermined portion of the component, and detects the taken enclose the outer contact rectangle all of the plurality of image portions corresponding to the component, the circumscribed rectangle of the candidate components plurality In some cases, one circumscribed rectangle of the component is detected based on a detection condition other than the dimension of the component, and the detection condition is that the area of the circumscribed rectangle among a plurality of candidates of the circumscribed rectangle of the component is Is the smallest, the sum of the areas of each image part contained in the circumscribed rectangle is maximized, the number of each image part contained in the circumscribed rectangle is maximized, or the circumscribed rectangle relative to the area of the circumscribed rectangle Each contained within the rectangle Of one of the plurality of requirements ratio of the sum of the area of the image areas is maximum, the circumscribed rectangle which satisfies at least one requirement, a circumscribed rectangle of the part.

In the component recognition apparatus according to the first aspect, as described above, one circumscribed rectangle that surrounds all of the plurality of image portions corresponding to the component is obtained by comparing the image recognition unit with the dimension data of the predetermined portion of the component. By configuring so as to detect, it is possible to recognize an image of a component made up of a plurality of image portions as one component (a circumscribed rectangle) without individually recognizing the plurality of image portions. In addition, the image recognition unit is configured to detect one circumscribed rectangle that surrounds all of the plurality of image portions corresponding to the component while comparing with the dimension data of the predetermined portion of the component, thereby The component can be recognized only by the dimension data of the image-recognizable portion of the component to be recognized without individually registering information that becomes the feature point such as the arrangement and shape of each component. In particular, when the shape of a part is complicated, it is necessary to increase the number of feature points. On the other hand, since the dimension data of a part already exists for each part, a plurality of setting points are not required. An image of a component made up of image portions can be recognized as one component. In the component recognition apparatus according to the first aspect, preferably, the image recognition unit performs image recognition based on a portion corresponding to noise in addition to a plurality of image portions corresponding to the component, and also corresponds to the component. One circumscribed rectangle that surrounds all of the plurality of image portions corresponding to the component is detected without surrounding the portion corresponding to the noise that is outside of the plurality of image portions.

  In the component recognition apparatus according to the first aspect described above, preferably, the image recognition unit circumscribes the plurality of image portions, and the dimension in one direction and the other direction orthogonal to the one direction has one direction of the predetermined portion of the component and By detecting a circumscribed rectangle that falls within a predetermined tolerance in comparison with the dimension data in the other direction, one circumscribed rectangle that surrounds all of the plurality of image portions corresponding to the part is detected. With this configuration, the circumscribed rectangle can be easily detected by comparing the dimensions of the predetermined part of the part with the dimensions in one direction and the other direction, and by detecting the circumscribed rectangle that falls within a predetermined tolerance. The parts can be recognized with high accuracy.

  In the component recognition apparatus according to the first aspect described above, preferably, when there are a plurality of candidate circumscribed rectangles for the component, the image recognizing unit detects the circumscribed rectangle of the component based on a detection condition other than the dimension of the component. The detection condition includes at least one of a detection condition relating to at least one area of the circumscribed rectangle or the image portion included in the circumscribed rectangle, or a detection condition relating to the number of image portions included in the circumscribed rectangle. With this configuration, even when a plurality of circumscribed rectangle candidates that match or approximate the dimensions of the component are detected, the circumscribed rectangle of the component can be detected with high accuracy based on the above-described detection conditions. In addition, the detection condition is configured to include at least one of a detection condition regarding at least one area of the circumscribed rectangle or the image portion included in the circumscribed rectangle, or a detection condition regarding the number of image portions included in the circumscribed rectangle. Thus, the circumscribed rectangle of the component can be detected without the need for information regarding the arrangement and position of the feature points of the component and other additional information.

  In the configuration for detecting a circumscribed rectangle of a component based on a detection condition other than the dimension of the component, preferably, the detection condition is a circumscribed rectangle that minimizes the area of the circumscribed rectangle among a plurality of candidates for the circumscribed rectangle of the component. The circumscribed rectangle of the part. With this configuration, for example, when there is noise (unnecessary light emitting points reflected in the captured image) in the vicinity of the component to be recognized in the captured image, the noise is larger than the size of the component by the amount of noise. When the size of the circumscribed rectangle including the size of the circumscribed rectangle increases, the circumscribed rectangle that matches the size of the part is detected from a plurality of candidates based on the detection condition that the circumscribed rectangle that minimizes the area of the circumscribed rectangle is the circumscribed rectangle of the part. be able to.

  In the configuration for detecting a circumscribed rectangle of a component based on a detection condition other than the size of the component, preferably, the detection condition is an area of each image portion included in the circumscribed rectangle among a plurality of candidates for the circumscribed rectangle of the component. The circumscribing rectangle that maximizes the total of the parts is defined as the circumscribing rectangle of the component. With this configuration, for example, when there is noise at a position in the vicinity of the component to be recognized in the captured image, the circumscribed rectangle including the noise surrounds all of the image parts constituting the component. If the total area of the image parts in the rectangle is smaller than the number of the circumscribed rectangles that make the total area of the image parts included in the circumscribed rectangle maximum From these candidates, a circumscribed rectangle that matches the dimensions of the part can be detected.

  In the configuration for detecting a circumscribed rectangle of a component based on a detection condition other than the size of the component, preferably, the detection condition is the number of image parts included in the circumscribed rectangle among a plurality of candidates for the circumscribed rectangle of the component. The circumscribing rectangle that maximizes is the circumscribing rectangle of the component. With this configuration, for example, when there is a small image part (part of a part) in the outer peripheral part of the recognition target part in the captured image, a circumscribed rectangle that surrounds all of the image parts constituting the part However, when the number of image parts in the rectangle is smaller than the circumscribed rectangle that does not contain a small image part (part of the part), the circumscribed rectangle that maximizes the number of each image part contained in the circumscribed rectangle is the part. Based on the detection condition of the circumscribed rectangle, a circumscribed rectangle that matches the dimensions of the component can be detected from a plurality of candidates.

  In the configuration for detecting the circumscribed rectangle of the component based on the detection condition other than the dimension of the component, preferably, the detection condition is included in the circumscribed rectangle with respect to the area of the circumscribed rectangle among a plurality of candidates for the circumscribed rectangle of the component. The circumscribed rectangle that maximizes the ratio of the total area of each image portion is defined as the circumscribed rectangle of the component. With this configuration, for example, when noise exists in the vicinity of the recognition target component in the captured image, the noise is detected including this noise, so that the noise is larger than the component size by the amount of noise. In the case where the size of the circumscribed rectangle including the size of the circumscribed rectangle increases while the ratio of the image portion occupying the area of the circumscribed rectangle including noise is relatively small, the area of each image portion included in the circumscribed rectangle with respect to the area of the circumscribed rectangle A circumscribed rectangle that does not include noise can be detected from a plurality of candidates based on a detection condition in which a circumscribed rectangle having the maximum total ratio is a circumscribed rectangle of a component.

  In the configuration for detecting the circumscribed rectangle of the component based on the detection condition other than the dimension of the component, preferably, the image recognition unit determines at least one detection condition to be applied when detecting the circumscribed rectangle of the component according to the type of the component. Are configured to be selectable. If configured in this way, the state of the captured image varies depending on the type of the component, such as a component including a plurality of small image portions in the captured image or a component in which noise is easily reflected depending on the shape and material of the component. In this case, it is possible to select an appropriate detection condition according to the type of part, such as which detection condition is applied when detecting a circumscribed rectangle, whether a single detection condition is applied, or whether multiple detection conditions are used in combination. By doing so, the circumscribed rectangle of the component can be detected with higher accuracy.

A component transfer apparatus according to a second aspect of the present invention includes a suction nozzle for sucking and holding a component, a head unit for transferring the sucked component to a predetermined transfer position, and the suction unit being sucked by the suction nozzle. An image of a component recognition unit, comprising: an imaging unit that captures a component; and a component recognition unit that includes an image recognition unit that performs image recognition based on a captured image including a plurality of image portions corresponding to the component captured by the imaging unit. recognition unit, in comparison with the dimension data of a predetermined portion of the component, and detects the taken enclose the outer contact rectangle all of the plurality of image portions corresponding to the component, if the circumscribed rectangle of the candidate of the component there are a plurality , One circumscribed rectangle of the component is detected based on a detection condition other than the dimension of the component, and the detection condition minimizes the area of the circumscribed rectangle among a plurality of candidate circumscribed rectangles of the component Or circumscribed rectangle Each image part included in the circumscribed rectangle relative to the area of the circumscribed rectangle, or the total of the areas of the image parts included in is maximized, or the number of each image part included in the circumscribed rectangle is maximized The circumscribed rectangle that satisfies at least one of the requirements for the maximum ratio of the total area is the circumscribed rectangle of the component, and the head unit corrects the position of the suction nozzle based on the image recognition result Thus, the component is transferred to a predetermined transfer position.

  In the component transfer apparatus according to the second aspect, as described above, one circumscribed rectangle that surrounds all of the plurality of image portions corresponding to the component while comparing the image recognition unit with the dimension data of the predetermined portion of the component. With this configuration, it is possible to recognize an image of a component composed of a plurality of image portions as one component (a circumscribed rectangle) without individually recognizing the plurality of image portions. In addition, the image recognition unit is configured to detect one circumscribed rectangle that surrounds all of the plurality of image portions corresponding to the component while comparing with the dimension data of the predetermined portion of the component, thereby The component can be recognized only by the dimension data of the image-recognizable portion of the component to be recognized without individually registering information that becomes the feature point such as the arrangement and shape of each component. In particular, when the shape of a part is complicated, it is necessary to increase the number of feature points. On the other hand, since the dimension data of a part already exists for each part, a plurality of setting points are not required. An image of a component made up of image portions can be recognized as one component. In addition, by configuring the head unit so that the position of the suction nozzle is corrected based on the image recognition result and the component is transferred to a predetermined transfer position, the component recognized as a circumscribed rectangle can be accurately transferred to the transfer position. Can be transferred.

  In this case, it is preferable that the head unit further includes a nozzle rotating device that rotationally drives the suction nozzle, and corrects the position and rotation position of the suction nozzle based on the image recognition result so that the component is predetermined at a predetermined transfer position. It is configured to transfer in the direction. According to this configuration, based on the image recognition result, in addition to the transfer position of the component recognized as a circumscribed rectangle, the direction (direction) of the component can be corrected, so that a predetermined direction is set at a predetermined transfer position. Thus, the transfer can be performed with higher accuracy.

It is a top view which shows the whole structure of the surface mounter by one Embodiment of this invention. It is a side view which shows the whole structure of the surface mounter by one Embodiment shown in FIG. It is a block diagram in connection with control of the surface mounter by one Embodiment shown in FIG. It is a figure which shows the picked-up image of the components by the component imaging device of the surface mounter by one Embodiment shown in FIG. It is a flowchart for demonstrating the component recognition process of the surface mounter by one Embodiment of this invention. It is a flowchart for demonstrating the component recognition process of the surface mounter by one Embodiment of this invention. It is a figure for demonstrating the component recognition process of the surface mounter by one Embodiment of this invention. It is a figure for demonstrating the component recognition process of the surface mounter by one Embodiment of this invention. It is a figure for demonstrating the component recognition process of the surface mounter by one Embodiment of this invention. It is a figure for demonstrating the component recognition process of the surface mounter by one Embodiment of this invention. It is a figure for demonstrating the component recognition process of the surface mounter by one Embodiment of this invention. It is a figure for demonstrating the 1st detection conditions for the circumscribed rectangle detection of the surface mounter by one Embodiment of this invention. It is a figure for demonstrating the 2nd detection conditions for the circumscribed rectangle detection of the surface mounter by one Embodiment of this invention. It is a figure for demonstrating the 3rd detection conditions for the circumscribed rectangle detection of the surface mounter by one Embodiment of this invention. It is a figure for demonstrating the 4th detection conditions for the circumscribed rectangle detection of the surface mounter by one Embodiment of this invention. It is a figure for demonstrating the component recognition process of the surface mounter by one Embodiment of this invention.

  Hereinafter, embodiments of the present invention will be described with reference to the drawings.

  The structure of the surface mounter 100 according to the embodiment of the present invention will be described below with reference to FIGS. In the present embodiment, an example in which the component recognition apparatus of the present invention is applied to a component recognition unit of a surface mounter will be described. The surface mounter 100 is an example of the “component transfer device” in the present invention.

  As shown in FIGS. 1 and 2, the surface mounter 100 according to the present embodiment is an apparatus for mounting a component 120 (see FIG. 2) on a printed board 110. As shown in FIG. 1, the surface mounter 100 includes a pair of substrate transport conveyors 10 extending in the X direction and a head unit 20 that can move in the XY directions above the pair of substrate transport conveyors 10. A plurality of tape feeders 121 for supplying the components 120 are arranged on both sides of the pair of substrate transport conveyors 10. The head unit 20 has a function of acquiring the component 120 from the tape feeder 121 and mounting the component 120 on the printed circuit board 110 on the substrate transport conveyor 10. As shown in FIG. 2, the substrate transport conveyor 10 is installed on the base 1, and the head unit 20 is arranged above the base 1. Hereinafter, a specific structure of the surface mounter 100 will be described.

  The pair of board conveyors 10 are configured to carry the printed board 110 in the X direction, stop the printed board 110 at a predetermined mounting operation position, and hold the printed board 110 by a board holding device (not shown). .

  The tape feeder 121 holds a reel around which a tape holding a plurality of components 120 at a predetermined interval is wound. The tape feeder 121 is configured to supply the component 120 from the tip of the tape feeder 121 by feeding a tape that holds the component 120 by rotating a reel. The component 120 is a small electronic component such as an IC, a transistor, a capacitor, and a resistor.

  The head unit 20 is configured to be movable in the X direction along the head unit support portion 30 extending in the X direction. Specifically, as shown in FIG. 1, the head unit support portion 30 includes a ball screw shaft 31, a servo motor 32 that rotates the ball screw shaft 31, and an X-direction guide rail (not shown). The head unit 20 also has a ball nut 21 to which a ball screw shaft 31 is screwed. The head unit 20 is configured to move in the X direction with respect to the head unit support 30 when the ball screw shaft 31 is rotated by a servo motor 32. Further, the head unit support section 30 is configured to be movable in the Y direction along a pair of fixed rail sections 40 extending in the Y direction provided on the base 1 so as to straddle the substrate transport conveyor 10. Specifically, one fixed rail portion 40 has a guide rail 41 that supports both end portions of the head unit support portion 30 so as to be movable in the Y direction, and the other fixed rail portion 40 includes a guide rail 41, A ball screw shaft 42 extending in the Y direction and a servo motor 43 for rotating the ball screw shaft 42 are provided, and the head unit support portion 30 is provided with a ball nut 33 into which the ball screw shaft 42 is screwed. . The head unit support portion 30 is configured to move in the Y direction along the guide rail 41 when the ball screw shaft 42 is rotated by the servo motor 43. With such a configuration, the head unit 20 is configured to be able to move on the base 1 in the XY directions.

  The head unit 20 is provided with six suction nozzles 22 arranged in a row in the X direction so as to protrude downward. Each suction nozzle 22 is configured to be able to generate a negative pressure state at the tip thereof by a negative pressure generator (not shown). The suction nozzle 22 can suck and hold the component 120 supplied from the tape feeder 121 at the tip by this negative pressure. Each of the six suction nozzles 22 is configured to be able to switch generation and release of the negative pressure state individually.

  Each suction nozzle 22 is configured to be movable in the vertical direction (Z direction) with respect to the head unit 20 by an elevating device (see FIG. 3) such as a servo motor. The surface mounter 100 carries the component 120 and the like while the suction nozzle 22 is located at the raised position, and sucks the component 120 from the tape feeder 121 and prints the printed circuit board 110 while the suction nozzle 22 is located at the lowered position. It is configured to make an implementation. Further, the suction nozzle 22 is configured so that the suction nozzle 22 itself can rotate around its axis by a nozzle rotating device (see FIG. 3) such as a servo motor. Thereby, in the surface mounting machine 100, it is possible to adjust the attitude | position (direction in a horizontal surface) of the components 120 hold | maintained at the front-end | tip of a nozzle by rotating the adsorption nozzle 22 in the middle of conveying the components 120. FIG. .

  In addition, a component imaging device 50 that uses a line sensor for imaging the posture of the component 120 sucked by the suction nozzle 22 is attached to the head unit 20. As shown in FIG. 2, the component imaging device 50 is attached to the head unit 20 so as to be movable in the X direction (the direction in which the six suction nozzles 22 are arranged). Specifically, the head unit 20 is provided with a ball screw shaft 23 extending in the X direction and a servo motor 24 that rotates the ball screw shaft 23, and the ball screw shaft 23 is screwed into the component imaging device 50. A ball nut 51 is provided. The component imaging device 50 is configured to move in the X direction with respect to the head unit 20 when the ball screw shaft 23 is rotated by the servo motor 24. The component imaging device 50 is attached with the imaging direction facing upward so as to face the lower surface of the suction nozzle 22. As a result, the component imaging device 50 can sequentially image the lower surface of the component 120 held by the six suction nozzles 22 arranged side by side in the X direction on the head unit 20 from below. Further, by attaching the component imaging device 50 to the head unit 20, the component imaging device 50 is moved relative to the head unit 20 while moving the head unit 20 to the mounting position while the component 120 is held by the suction nozzle 22. It is possible to move and image the posture of the component 120 (the suction state to the suction nozzle 22). Based on the captured image of the component 120 imaged by this component imaging device 50, it is comprised so that component recognition mentioned later can be performed. The component imaging device 50 is an example of the “imaging unit” in the present invention.

  The operation of the surface mounter 100 is controlled by the control device 60 shown in FIG. The control device 60 includes a main calculation unit 61, an axis control unit 62, a storage unit 63, and an image processing unit 64. The control device 60 is an example of the “component recognition unit” and “component recognition device” in the present invention. The image processing unit 64 is an example of the “image recognition unit” in the present invention.

  The main arithmetic unit 61 is composed of a CPU for executing logical operations. The main calculation unit 61 is a lifting device (servo) for moving up and down the component imaging device 50 and each suction nozzle 22 via the axis control unit 62 and the image processing unit 64 according to a program stored in the ROM of the storage unit 63. Motor), a nozzle rotation device for rotation (servo motor, etc.), and other servo motors. At the time of mounting, the main calculation unit 61 includes the axis control unit 62 and the storage unit 63 so that the components 120 are sequentially mounted at predetermined mounting positions on the printed circuit board 110 according to the mounting program stored in the storage unit 63. The image processing unit 64 is configured to be controlled.

  Based on the control signal output from the main calculation unit 61, the axis control unit 62 performs servo motors of the surface mounter 100 (servo motors 43 for moving the head unit support unit 30 in the Y direction (see FIG. 1)). Servo motor 32 (see FIG. 1) for moving the head unit 20 in the X direction, servo motor (elevating device) of the lifting mechanism for moving the six suction nozzles 22 in the vertical direction, and six suctions Servo motor (nozzle rotation device) of the rotation mechanism for rotating the nozzle 22 in the R-axis direction (rotation direction around the central axis of each suction nozzle) and the component imaging device 50 for moving in the X direction The servomotor 24 (see FIG. 2)) and the like are controlled.

  The storage unit 63 includes a ROM (Read Only Memory) that stores a program for controlling the CPU, a RAM (Random Access Memory) that temporarily stores various data during the operation of the apparatus, and the like. In addition, the storage unit 63 stores a mounting program for manufacturing a predetermined printed circuit board 110, and is applied to the types, dimensions, and component recognitions of all the parts 120 mounted on the printed circuit board 110. Parts data such as a circumscribed rectangle detection condition, a mounting position, and an angle are set in advance. At the time of mounting, the component 120 sucked by the suction nozzle 22 is recognized based on the dimension data of the component 120 included in the component data. This dimension data is data indicating the dimensions of a circumscribed rectangular area in which an image can be recognized in a captured image 200 (see FIG. 4) in which the component 120 is captured by the component imaging device 50 and predetermined image processing described later is executed. is there. For example, as shown in FIG. 4, the component 120 is composed of a plurality of image portions 130 (regions without hatching) excluding the noise 140. At this time, a dimension A × B of a component circumscribing rectangle 150 (described later) surrounding the image portion 130 of the component 120 is stored in the storage unit 63 as dimension data of the component 120. Since the dimension data of the part 120 is supplied from the manufacturer of the part, the existing dimension data can be used without measuring the dimension of the part 120.

  The image processing unit 64 reads out the imaging signal from the component imaging device 50 at a predetermined timing based on the control signal output from the main calculation unit 61, and performs predetermined image processing on the read imaging signal. The image data suitable for recognizing the component 120 is generated. In the present embodiment, the image processing unit 64 is configured to perform component recognition of the component 120 based on the captured image 200 of the component 120 captured by the component imaging device 50. Specifically, as illustrated in FIG. 4, the image processing unit 64 surrounds the entire image portion 130 corresponding to the component 120 based on the captured image 200 of the component 120 while comparing with the dimension data of the component 120. One component circumscribed rectangle 150 is configured to be detected. Component recognition is performed by detecting a circumscribed rectangle (component circumscribed rectangle 150) of the component 120 that matches the dimension (A × B) of the component 120 within a predetermined tolerance. That is, the component recognition is performed by detecting the component circumscribed rectangle 150 having a dimension of A × B corresponding to the dimension data of the predetermined part of the component 120 from the captured image 200 by the component recognition process. Thus, the component recognition is performed on the recognition target component 120 based on the dimension A × B of the predetermined portion capable of image recognition without requiring detailed information regarding the arrangement and shape of each part of the component 120. It is possible.

  In this embodiment, the image processing unit 64 is configured to detect the component circumscribing rectangle 150 based on detection conditions other than the dimension A × B of the component 120 when there are a plurality of candidates for the component circumscribing rectangle 150. ing. This detection condition includes a plurality of different detection conditions, and is configured such that at least one detection condition applied according to the type of the component 120 is selected. Of the plurality of detection conditions, which detection condition is applied to the recognition target component 120 is registered in advance in the component data of the storage unit 63. Therefore, when there are a plurality of candidate parts circumscribing rectangles 150 at the time of component recognition, the component data of the component 120 to be recognized is read from the storage unit 63, and the detection condition corresponding to the type of the component 120 is applied. It is configured. This detection condition can also be applied in combination with a plurality of detection conditions. The detection condition is set so as to include at least one of a detection condition regarding at least one area of the circumscribed rectangle or the image portion 130 included in the circumscribed rectangle or a detection condition regarding the number of image portions 130 included in the circumscribed rectangle. Yes.

  FIG. 5 and FIG. 6 are flowcharts for explaining processing when performing component recognition based on the captured image of the captured component. 7 to 16 are diagrams for explaining the content of the component recognition processing. Next, a component recognition process of the surface mounter 100 according to the embodiment of the present invention will be described in detail with reference to FIGS.

  In this embodiment, as shown in FIGS. 1 and 2, when the component 120 is sucked from the tape feeder 121 by the suction nozzle 22, the head unit 20 moves to a predetermined mounting position on the printed circuit board 110, and the suction nozzle 22 The lower surface of the component 120 sucked by 22 is imaged by the component imaging device 50. At this time, as shown in FIG. 4, the captured image 200 of the component 120 is read from the component imaging device 50 by the image processing unit 64 and the dimension data of the component 120 stored in the storage unit 63 is read. Based on the captured image 200 of the component 120, a component circumscribing rectangle 150 surrounding all of the plurality of image portions 130 corresponding to the component 120 is detected while comparing with the dimension data of the component 120. In addition, as shown in FIG. 7, the detection of the component circumscribed rectangle 150 is repeatedly performed a plurality of times by rotating the angle of the coordinate system by an angle θ within a predetermined angle range. Thereby, even when the component 120 is imaged in a state tilted with respect to the component imaging device 50, the component circumscribing rectangle 150 of the component 120 can be correctly recognized. In the following description, a circumscribed rectangle that surrounds all of the plurality of image portions 130 of the component 120 and has the set dimension A × B is represented as a “component circumscribed rectangle 150”. On the other hand, when simply representing “the circumscribed rectangle”, all circumscribed rectangles that are detected in the process of the component recognition process and can be candidates for the component circumscribed rectangle 150 that is the target of component recognition are included. That is, component recognition is performed by selecting a circumscribed rectangle that matches the component circumscribed rectangle 150 from a plurality of circumscribed rectangles (candidates) detected by the following processing.

  As shown in FIG. 5, in step S <b> 1, predetermined image processing is performed on the captured image 200 of the component 120 read from the component imaging device 50 by the image processing unit 64, and each image portion 130 on the captured image 200 is processed. Contour detection is performed. As shown in FIG. 4, by this contour detection, the coordinates of the contour of the noise 140 reflected on each image portion 130 and the captured image 200 in the coordinate system XY are acquired. Here, the coordinate system XY is a coordinate system set in advance in the captured image 200 by the component imaging device 50. In contour detection, first, a binary image of a captured image is obtained by performing binarization based on a predetermined threshold relating to the density of a captured image (not shown) having light and shade. The captured image 200 shown in FIG. 4 shows the obtained binary image. Next, contour tracking of the captured image 200 which is a binary image is performed to detect a chain code. That is, an arbitrary boundary pixel (a pixel at the boundary between 0 (white) and 1 (black)) is detected in the binary image, and the boundary pixel is selected from the eight pixels surrounding the boundary pixel (center boundary pixel). A certain pixel (peripheral boundary pixel) is detected, and the direction of the peripheral boundary pixel with respect to the central boundary pixel is generated as a chain code. After that, by repeating the generation of the chain code with the peripheral boundary pixel detected in the chain code direction as the central boundary pixel, the contour of the image portion 130 that is an area surrounded by the boundary pixel is detected. Thereby, the coordinates of the contour of each image portion 130 (noise 140) of the captured image 200 are acquired. In the present embodiment, for convenience of explanation, in the captured image 200, the component portion of the component 120 is described as the image portion 130, and an unnecessary light emission point that does not configure the component 120 is described as noise 140. Since both are not processed, both are processed as a light emitting region.

In step S2, the maximum and minimum coordinates in the XY coordinate system of each image portion 130 included in the captured image 200 are acquired by the image processing unit 64 based on the contour coordinates of each image portion 130 detected in step S1. The For example, for the image portion 130 located on the lower left side in FIG. 4, the maximum coordinates x _max and y _max and the minimum coordinates x _min and y _min are acquired. The detection of the maximum coordinate and the minimum coordinate is performed for all the image portions 130 (noise 140). In addition, the maximum coordinates and minimum coordinates of all the acquired image portions 130 (noise 140) are arranged in the order of ascending values (ascending order) for each of the X coordinates and the Y coordinates, and the maximum coordinates are values. They are arranged in the descending order (descending order) and stored in the storage unit 63.

  In step S3, the difference between the minimum X coordinate and the maximum X coordinate for the acquired maximum X coordinate and minimum X coordinate of each image portion 130 is obtained, and compared with the dimension A of the component 120 in the X coordinate direction. A combination of the maximum X coordinate and the minimum X coordinate that matches the dimension A in the X coordinate direction of 120 is detected. The image processing unit 64 sequentially reads out the minimum X coordinate arranged in ascending order and the maximum X coordinate arranged in descending order from the storage unit 63, and compares the difference between the combination of the minimum X coordinate and the maximum X coordinate. Done. The comparison of whether or not the dimension 120 of the part 120 coincides with the dimension A in the X coordinate direction is based on the difference between the calculated minimum X coordinate and the maximum X coordinate (the size of the line segment) being a predetermined tolerance of a preset dimension A. If it is within the tolerance, it is determined that it matches the dimension A. If there is a combination that matches the dimension A of the component 120 and the matching combination is not a combination of the maximum X coordinate and the minimum X coordinate that is already stored in the storage unit 63, the combination is newly stored in the storage unit 63. Remember. If the matching combination is the same as the combination of the maximum X coordinate and the minimum X coordinate already stored in the storage unit 63, another combination that matches the dimension A of the part 120 and that is not stored in the storage unit 63. The combination of the maximum X coordinate and the minimum X coordinate is obtained.

As shown in FIG. 8, a minimum X-coordinate _{x 1} of the image portion 130a, a difference A1 between the maximum X-coordinate _{x 2} of the image portion 130b is substantially coincident with the dimension A of the component 120, and within the tolerances set Therefore, the combination of the minimum X coordinate x ₁ and the maximum X coordinate x ₂ is not the combination of the maximum X coordinate and the minimum X coordinate already stored in the storage unit 63 and coincides with the dimension A of the component 120. It is newly memorized as a combination. On the other hand, a minimum X-coordinate _{x 1} of the image portion 130a, the magnitude of the difference A2 between the maximum X-coordinate _{x 3} of the image portion 130c is smaller than the dimension A of the component 120. For this reason, the difference A2 does not fall within the set tolerance range and does not coincide with the dimension A of the part 120. Therefore, the difference A2 is excluded from the combination that coincides with the dimension A of the part 120 and is not detected.

  In step S <b> 4, the image processing unit 64 determines whether or not there is a combination of the maximum X coordinate and the minimum X coordinate within the predetermined tolerance range of the dimension A of the component 120 in the X coordinate direction. That is, it is determined whether or not there is a combination of the maximum X coordinate and the minimum X coordinate newly stored in step S3. If there is a combination, the process proceeds to step S5. If there is no combination, there is no combination that matches the dimension A of the part 120, or even if there is a combination that has already been selected in the preceding step S3 and has already been performed in step S5 and later. Yes, the process proceeds to step S10.

  In step S5, the maximum Y coordinate and the maximum Y coordinate of each image portion 130 within the X coordinate range determined by the combination of the minimum X coordinate and the maximum X coordinate newly detected in step S3 and stored in the storage unit 63 by the image processing unit 64. For the minimum Y coordinate, the difference between the minimum Y coordinate and the maximum Y coordinate is determined. The difference between the minimum Y coordinate and the maximum Y coordinate is compared with the dimension B of the part 120 in the Y coordinate direction, so that the maximum Y coordinate and the minimum Y coordinate corresponding to the dimension B of the part 120 in the Y coordinate direction are compared. A combination is detected. At this time, the minimum Y coordinate arranged in the ascending order and the maximum Y coordinate arranged in the descending order are sequentially read from the storage unit 63, and the difference is compared for the combination of the minimum Y coordinate and the minimum Y coordinate. The comparison of whether or not the dimension 120 matches the dimension B in the Y-coordinate direction of the component 120 is based on the difference between the obtained minimum Y coordinate and maximum Y coordinate (size of the line segment) being a predetermined tolerance of the dimension B set in advance. If it is within the intersection, it is determined that it matches the dimension B. If there is a combination that matches the dimension B of the component 120 and the matching combination is not a combination of the maximum Y coordinate and the minimum Y coordinate already stored in the storage unit 63, the combination is newly stored in the storage unit 63. Remember. If the matching combination is the same as the combination of the maximum Y coordinate and the minimum Y coordinate already stored in the storage unit 63, another match is made with the dimension B of the component 120 and not stored in the storage unit 63. The combination of the maximum Y coordinate and the minimum Y coordinate is obtained.

As shown in FIG. 9, in the range of a difference A1 between the minimum X-coordinate x ₁ and the maximum X-coordinate x ₂ detected, when compared with respect to the Y coordinate direction, the smallest Y-coordinate y ₁ of the image portion 130d , the difference B1 between the maximum Y-coordinate _{y 2} of the image portion 130b is a part 120 Y-coordinate direction dimension B and substantially coincident, since the range of tolerance set, the coordinate _{y 1} and the coordinate _{y 2} The combination is newly stored as a combination that matches the dimension B of the component 120 when the combination is not a combination of the maximum Y coordinate and the minimum Y coordinate already stored in the storage unit 63. On the other hand, the minimum Y-coordinate _{y 3} of the image portion 130e, the magnitude of the difference B2 between the maximum Y-coordinate _{y 2} of the image portion 130b is smaller than the dimensions of the part 120 B. For this reason, the difference B2 does not fall within the set tolerance range and does not coincide with the dimension B of the part 120. Therefore, the difference B2 is excluded from the combination that coincides with the dimension B of the part 120 and is not detected. Note that the combination of the minimum Y coordinate y ₄ of the noise 140a and the maximum Y coordinate y ₅ of the image portion 130e (130g) also has a difference B3 between the minimum Y coordinate y ₄ and the maximum Y coordinate y ₅ to the dimension B of the component 120. If within a predefined cross is may be detected as matching the dimensions of the parts 120 B, the combination of the minimum Y-coordinate y ₄ and a maximum Y-coordinate y _5, the minimum Y-coordinate y ₁ and the maximum Y If the combination of coordinates y ₂ is already stored in the storage unit 63, or steps S6, S7 and S8 described later are sequentially performed using the combination of the minimum Y coordinate y ₁ and the maximum Y coordinate y ₂ , It becomes No in step S8 and is detected when step S5 is performed again, and is newly stored as a combination that matches the dimension B of the component 120.

  In step S <b> 6, the image processing unit 64 determines whether or not there is a combination of the maximum Y coordinate and the minimum Y coordinate that falls within a predetermined tolerance range of the dimension B of the component 120 in the Y direction. That is, it is determined whether or not there is a combination of the maximum Y coordinate and the minimum Y coordinate newly stored in step S5, and if it exists, the process proceeds to step S7. If there is no combination, there is no combination that matches the dimension B of the part 120, or even if there is a combination that has already been selected in the preceding step S5 and has already been performed in step S7 and later. Return to step S3.

In the example shown in FIG. 9, the combination of the maximum Y coordinate and the minimum Y coordinate within the range of the difference A1 between the minimum X-coordinate x ₁ and the maximum X-coordinate x _2, which is detected to coincide with the dimensions of the component 120 B is present Thus, a circumscribed rectangle 151 having a size of A1 × B1 and surrounding nine image portions 130 (130a and the like) is detected. Note that a circumscribed rectangle 152 having a size of A1 × B3 and surrounding six image portions 130 (such as 130e) and two noises 140 (such as 140a) can be detected. Since step S7 and subsequent steps using 151 are already performed, the detection is performed through steps S5 and S6 which are performed again.

  When a circumscribed rectangle having a dimension that matches or approximates the dimension A × B of the part 120 and whose opposite sides are parallel to the X coordinate axis and the Y coordinate axis is obtained, the process proceeds to step S7, and these are obtained. The group of image portions 130 (noise 140) included in the circumscribed rectangle is rotated by a predetermined angle with respect to the coordinate system XY, and the opposite sides are parallel to the X coordinate axis and the Y coordinate axis, respectively, at the rotated angles. A circumscribed rectangle made up of rectangles is obtained, and an angle at which a circumscribed rectangle that minimizes the inner area of the plurality of circumscribed rectangles is obtained. This is because when the part 120 is contained in the circumscribed rectangle, the dimension A × B of the part 120 set in advance is coincident with the rotation angle at which the rectangular area is minimized. As shown in FIG. 4, the dimension A × B of the part 120 is a dimension of the part circumscribed rectangle 150 when the part 120 is imaged without being inclined with respect to the coordinate system XY. FIG. 7 shows the X1 axis and Y1 axis of the coordinate system in which the opposite sides are rotated when obtaining the circumscribed rectangle with respect to the image portion 130 with the image portion 130 fixed from the state of FIG. Circumscribed rectangle consisting of a rectangle parallel to the X2 axis Y2 axis of the coordinate system with opposite sides rotated, and circumscribed rectangle consisting of a rectangle parallel to the X3 axis Y3 axis of the coordinate system with opposite sides rotated respectively The state which calculated | required the rectangle is illustrated. It can be seen that the circumscribed rectangle increases as the rotation angle increases. The component imaging device 50 is disposed in the head unit 20 so as to be movable in the X direction so that the coordinate system XY of the captured image 200 matches the coordinate system XY set on the base 1. Since the minimum circumscribed rectangle coincides with the part 120 in FIG. 7, the circumscribed rectangle obtained by the coordinate system XY coincides with the part 120, so that the coordinate system X 1 -Y 1 is the part 120. If it is assumed that the coordinate system XY on the base 1 coincides with the coordinate system XY on the base 1 (assuming that it matches the coordinate system X1-Y1), the component 120 is It can be seen that it is tilted clockwise by θ. Similarly, when it is assumed that the coordinate system X3-Y3 coincides with the coordinate system XY on the base 1 when the part 120 is imaged, the part 120 is a timepiece with respect to the coordinate system XY on the base 1. It can be seen that the direction is inclined by 3θ.

  Here, as shown in FIG. 10, a circumscribed rectangle 151 having a dimension that matches or approximates the dimension A × B of the component 120 is obtained by the processing from steps S2 to S6, and after step S7 to S8, step S5 is performed again. , The circumscribed rectangle 152 having a dimension that matches or approximates the dimension A × B of the part 120 is obtained by the process of S6, and the circumscribed rectangle 153 is obtained through the steps S7 to S8 after the circumscribed rectangle 152 is obtained. S6, and further returning to step S3, suppose that it was obtained by the processing of steps S3 to S6. The circumscribed rectangle 153 includes image portions 130d, 130e, 130f, and 130g, and noises 140b and 140c. At this time, the circumscribed rectangle 153 has a dimension of A3 × B4. In step S7 after the circumscribed rectangle 153 is obtained, the image portions 130d, 130e, 130f and 130g included in the circumscribed rectangle 153 and the noises 140b and 140c are rotated to detect the angle at which the area of the circumscribed rectangle is minimized. To do. The circumscribed rectangle 153 in the state before the rotation in FIG. 11 has a minimum area when the circumscribed rectangle 153a shown in the state after the rotation in FIG. In this case, the dimension of the circumscribed rectangle 153a is A4 × B5. The dimension B5 of the circumscribed rectangle 153a in the Y coordinate direction does not match the dimension B of the component 120 in the Y coordinate direction. In this way, the rotation angle φ that minimizes the area of the circumscribed rectangle is obtained. The principle for obtaining the rotation angle φ is as described above with reference to FIG.

  Next, in step S8, the image processing unit 64 determines whether or not the dimension of the circumscribed rectangle at the rotation angle φ that minimizes the area matches or approximates the dimension A × B of the component 120. When the dimension of the circumscribed rectangle matches or approximates the dimension A × B of the part 120, the process proceeds to step S9. In step S9, it is referred to whether the combination of the maximum X coordinate and minimum X coordinate and the maximum Y coordinate and minimum Y coordinate forming this circumscribed rectangle is already registered in the storage unit 63. The part circumscribing rectangle 150 of the part 120 is registered as a candidate. Further, when the dimension A4 × B5 does not coincide with the dimension A × B of the part 120 as in the circumscribed rectangle 153a in the state after rotation in FIG. 11, the circumscribed rectangle 153a is excluded, and the process proceeds to step S5. The detection of the combination of the image portions 130 whose dimensions in the Y coordinate direction coincide with each other is attempted again.

  On the other hand, if the combination of the maximum X coordinate and the minimum X coordinate detected in step S3 does not coincide with the dimension A of the part 120 within a predetermined tolerance range, or is matched by repeating steps S3 to S9. When there is no new combination to be performed, it is determined in step S4 that there is no combination that matches the dimension A of the part 120 and there is no new combination, and the process proceeds to step S10.

  In step S10, it is determined whether or not the circumscribed rectangle has been detected in the angle range specified in advance. If the component circumscribed rectangle 150 is not detected in the specified angle range, the process proceeds to step S11. In step S11, the coordinate system XY is rotated by a predetermined angle θ, and the detection processing from steps S2 to S9 is performed again. That is, as shown in FIG. 7, the circumscribed rectangle detection process from steps S2 to S9 is repeatedly performed a plurality of times by a predetermined angle range by rotating the coordinate system XY of the component imaging device 50 by the angle θ. . Here, in this embodiment, the rotation angle θ of this coordinate system is 5 degrees. Further, the predetermined angle range in which the detection is repeated is performed in a range of −45 degrees to +45 degrees. For this reason, first, the detection processing from steps S2 to S9 performed in the coordinate system XY of the component imaging apparatus 50 is performed by the coordinate system X1-Y1 rotated by the angle θ. When detection and coordinate rotation are repeated so as to be coordinate system X2-Y2 and coordinate system X3-Y3, and detection processing is performed for a predetermined angle range, it is determined that detection for a specified angle range is performed in step S10. The process proceeds to step S12. Note that the detection of the component circumscribing rectangle 150 in each rotated coordinate system is performed by repeating the above steps S2 to S9, and thus description thereof is omitted. If a circumscribed rectangle having a dimension that matches or approximates the dimension A × B of the part 120 is detected in the rotated coordinate systems X1-Y1, X2-Y2, and X3-Y3, the circumscribed rectangle is detected in step S9. The combination of the maximum X coordinate and the minimum X coordinate forming the rectangle and the maximum Y coordinate and the minimum Y coordinate is registered or referred to in the storage unit 63. If the combination is not registered, the component circumscribing of the component 120 is performed. The rectangle 150 is registered as a candidate.

  In addition, the coordinate system XY of the component imaging device 50 is set so as to coincide with the coordinate system XY set on the base 1, and the captured image 200 is obtained with respect to the captured image 200 obtained when the component 120 is captured. Coordinate system XY coincides with the coordinate system XY set on the base 1. By obtaining the circumscribed rectangle using this coordinate system, the minimum circumscribed rectangle is not necessarily obtained, but from the coordinate system XY of the captured image 200 that matches the coordinate system XY set on the base 1. Since the coordinate system rotated counterclockwise, for example, the coordinate system of X1-Y1, X2-Y2, or X3-Y3 shown in FIG. Step S11 is performed. The position and inclination (rotation angle φ) of the component 120 in each coordinate system obtained using the coordinate system of X1-Y1, X2-Y2, or X3-Y3 is the same as the coordinate system XY set on the base 1 Coordinate conversion is performed to the coordinate system XY of the captured image 200 that coincides, and the position and inclination of the component 120 in the coordinate system XY of the captured image 200 are calculated.

  In step S <b> 12 of FIG. 6, the image processing unit 64 refers to the storage unit 63 and determines whether or not a plurality of circumscribed rectangles that are candidates for the component circumscribed rectangle 150 of the component 120 are detected. As described above, when circumscribing coordinates are detected a plurality of times within the specified angle range, a plurality of circumscribed rectangles having dimensions that substantially match the dimension A × B of the component 120 are detected, and a plurality of combinations are candidates. It may be registered as. In this case, it is determined that there are a plurality of candidates, and the process proceeds to step S13. On the other hand, if there is only one candidate for the component circumscribed rectangle 150 detected, the process proceeds to step S15. The detected circumscribed rectangle is recognized as the component circumscribed rectangle 150 of the component 120, and the component recognition result is output. This output includes the position and inclination (rotation angle) in the XY direction with respect to the center of the suction nozzle 22 of the component circumscribing rectangle 150 of the component 120 in the coordinate system XY of the captured image 200 that matches the coordinate system XY set on the base 1. φ), including outer shape.

  In step S13, the search condition included in the component data is read from the storage unit 63. This search condition is a condition for selecting the component circumscribing rectangle 150 of one component 120 from a plurality of candidates of the component circumscribing rectangle 150. The shape of the component 120 varies depending on the type of the component 120, and includes various components such as the component 120 in which the noise 140 is easily reflected in the captured image 200, and the component 120 having a plurality of terminals and including many small image portions 130 in the captured image. It is. For this reason, the detection condition applied to the component 120 to be detected is selected based on the component data of the imaged component 120.

  In step S14, the component circumscribing rectangle 150 of the component 120 is detected from the plurality of candidates based on the selected detection condition. Here, the four detection conditions applied will be described. In the following, a case where one circumscribed rectangle is selected from two candidates of the component circumscribed rectangle 150 will be described as a typical example.

  First, the first detection condition is that a circumscribed rectangle having the maximum total area of each image portion included in the circumscribed rectangle is selected from among a plurality of candidates of the component circumscribed rectangle 150 of the component 120. Is a detection condition. As shown in FIG. 12, when a captured image 201 includes a component 120a composed of 12 image portions 131, a circumscribed rectangle 154 including these 12 image portions 131 is a component of the set component 120a. It coincides with the circumscribed rectangle 150. At this time, since three noises 141 are present at the same interval on the opposite side (right side) of the image portion 131 on the left side of the image, two candidates (circumscribed rectangles 154 and 155) having substantially the same dimensions are detected. Suppose. Here, the noise that appears in the captured image is usually a light emission point that is smaller than that of the image portion constituting the component 120. Therefore, the circumscribed rectangle having the maximum total area of the image portion 131 (noise 141) included in the rectangle is recognized as the component circumscribed rectangle 150 of the component 120a. As a result, as shown in FIG. 12, the circumscribed rectangle 154 including the twelve image portions 131 is included in the rectangle rather than the circumscribed rectangle 155 including the nine image portions 131 and the three noises 141. Since the total area of the parts becomes large, the circumscribed rectangle 154 is selected as the circumscribed rectangle of the component 120a. As a result, it is possible to correctly recognize the component circumscribing rectangle 150 of the component 120 in which the area of each image portion is larger than the area of the noise 141.

  Next, the second detection condition is that the circumscribing rectangle in which the number of each image portion 130 included in the circumscribing rectangle is the largest among the plurality of candidates of the component circumscribing rectangle 150 of the component 120 is the component circumscribing rectangle 150 of the component 120. Is a detection condition. As shown in FIG. 13, when a captured image 202 includes a part 120b composed of twelve large image portions 132a and three small image portions 132b, these twelve image portions 132a and A circumscribed rectangle 156 including three small image portions 132b matches the component circumscribed rectangle 150 set for the component 120b. At this time, a circumscribed rectangle 157 including only twelve image portions 132a other than the three small image portions 132b, twelve large image portions 132a, and three small image portions 132b are included. Assume that the circumscribed rectangle 156 is detected as a candidate. In this case, the circumscribed rectangle 156 including a total of 15 image portions 132a and 132b is more than the circumscribed rectangle 157 including a total of 12 image portions 132a. Due to the large number, the circumscribed rectangle 156 is selected as the circumscribed rectangle of the component 120b. As a result, it is possible to correctly recognize the component circumscribing rectangle 150 of the component 120 in which many small image portions are included in the captured image.

  Next, the third detection condition is a detection condition in which a circumscribed rectangle having the smallest area of the circumscribed rectangle among the plurality of candidates of the component circumscribed rectangle 150 of the component 120 is set as the component circumscribed rectangle 150 of the component 120. As shown in FIG. 14, when a captured image 203 includes a component 120c including 12 image portions 133, a circumscribed rectangle 158 including these 12 image portions 133 is set as the component 120c. It coincides with the circumscribed rectangle 150. At this time, since the noise 142 on the lower right side of the image exists, it is assumed that two candidates (the circumscribed rectangles 158 and 159) having substantially the same dimensions are detected. That is, a circumscribed rectangle 158 including 12 image portions 133 and a circumscribed rectangle 159 including 12 image portions 133 and one noise 142 are detected as candidates. The circumscribed rectangle 158 has a dimension of A5 × B6, and the circumscribed rectangle 159 has a dimension A6 × B7 larger than the circumscribed rectangle 158. In this case, the circumscribed rectangle 159 including the image portion (noise) 142 has a larger area than the circumscribed rectangle 158 by the amount circumscribing the noise 142 outside the circumscribed rectangle 158. Therefore, the circumscribed rectangle 158 having a smaller area is replaced by the component 120c. Is selected as a circumscribed rectangle. Accordingly, it is possible to correctly recognize the component circumscribing rectangle 150 of the component 120 in which noise is easily reflected in the vicinity of the component circumscribing rectangle 150 of the component 120 depending on the material of the component 120 and the like.

  Next, the fourth detection condition is that a circumscription that maximizes the ratio of the total area of each image portion included in the circumscribed rectangle to the area of the circumscribed rectangle among the plurality of candidates of the component circumscribed rectangle 150 of the component 120. This is a detection condition in which the rectangle is a component circumscribing rectangle 150 of the component 120. As shown in FIG. 15, when a captured image 204 includes a component 120d including 12 image portions 134, a circumscribed rectangle 160 including these 12 image portions 134 is set as the component 120d. It coincides with the circumscribed rectangle 150. At this time, since there are six noises 143 in the image, it is assumed that two candidates (circumscribed rectangles 160 and 161) having substantially the same dimensions are detected. That is, a circumscribed rectangle 160 including 12 image portions 134 and two noises 143 and a circumscribed rectangle 161 including eight image portions 134 and six noises 143 are detected as candidates. In this case, even if the total number of image parts included in the circumscribed rectangle is the same and the areas of the circumscribed rectangles 160 and 161 are substantially the same, the circumscribed rectangle 161 containing a large number of small image parts (noise) 143 is also included. The ratio of the total area of the image portions 134 and 143 included in the circumscribed rectangle 161 to the area of the circumscribed rectangle 161 is smaller than that of the circumscribed rectangle 160. Therefore, the circumscribed rectangle 160 is selected as the component circumscribed rectangle 150 of the component 120d.

  By selecting and applying one or more of the detection conditions as described above according to the type of the part 120, a plurality of candidate parts circumscribing rectangles 150 whose dimensions match or approximate to the part circumscribing rectangle 150 of the part 120 are obtained. Even when it exists, it is possible to correctly detect the component circumscribing rectangle 150.

  In step S14 shown in FIG. 6, the image processing unit 64 selects one of a plurality of candidates for the component circumscribing rectangle 150 of the component 120 based on the detection condition registered in advance in the component data according to the type of the component 120 to be recognized. The component circumscribed rectangle 150 is detected.

  In step S15, the component circumscribed rectangle 150 of the detected component 120 is output including the position and inclination with respect to the center of the suction nozzle 22 as a result of component recognition. Accordingly, as shown in FIG. 16, a component circumscribing rectangle 150 that surrounds the image portion 130 that forms the component 120 and matches the set dimension A × B is detected from the captured image 200. At the same time, the rotation angle ψ from which the component circumscribed rectangle 150 is obtained is acquired. As shown in FIG. 16, when the component 120 is imaged without tilting with respect to the component imaging device 50, the rotation angle ψ is 0 degree and is detected in the coordinate system XY of the component imaging device 50. The circumscribed rectangle is the component circumscribed rectangle 150 of the component 120.

As described above, the component recognition processing of the component 120 based on the captured image 200 is performed. Further, the posture of the component 120 sucked by the suction nozzle 22 is recognized based on the component recognition result. Further, for example, when the component 120 is mounted with the center position of the component 120 as a reference, as shown in FIG. 16, the coordinates (x _c , y _c ) is obtained. The mounting position of the component 120 on the printed circuit board 110 is corrected based on the center position C and the posture (inclination) of the sucked component 120.

  Next, the mounting operation of the component 120 on the printed circuit board 110 by the surface mounter 100 of the present embodiment will be described.

  First, as shown in FIG. 1, the printed circuit board 110 is carried onto the base 1 via the pair of board transport conveyors 10, and is transported to the mounting work position at the center of the base 1 and is fixedly held. At this time, the head unit 20 moves above the printed board 110 and images a plurality of board marks (not shown) on the printed board 110 by a board imaging device (not shown). Thereby, the reference point of the mounting position of the component is acquired.

  In parallel with the loading operation of the printed circuit board 110, the component 120 to be mounted is taken out from the tape feeder 121 by the head unit 20. Specifically, when the head unit 20 is moved above a predetermined component extraction position of the tape feeder 121, the suction nozzle 22 of the head unit 20 is placed above the component 120 to be mounted held by the tape feeder 121. Be placed.

  Thereafter, the suction nozzle 22 is lowered, and negative pressure is supplied to the tip of the suction nozzle 22 at a predetermined timing. Thereby, the component 120 on the tape feeder 121 is sucked and held by the suction nozzle 22. Thereafter, the tape feeder 121 feeds the tape until the next component 120 is arranged at the component removal position.

  After the component 120 is sucked, the suction nozzle 22 holding the component 120 is raised, and the head unit 20 is moved to the mounting position above the printed board 110 with the board mark as a reference. At this time, as shown in FIG. 2, while moving the head unit 20, the component imaging device 50 attached to the head unit 20 is moved in the X direction, thereby imaging the component 120 held by the suction nozzle 22. Done. As a result, an image of the lower surface of the component 120 is taken.

  At this time, the captured image 200 of the lower surface of the component 120 is read from the component imaging device 50 by the image processing unit 64, and the dimensions and detection conditions of the sucked component 120 are read from the storage unit 63. Then, the image processing unit 64 performs the above-described component recognition processing based on the captured image 200 of the lower surface of the component 120 and the dimension data of the component 120, and determines the position and orientation (rotation angle) of the captured component 120. recognize. Based on this component recognition result, the amount of deviation of the suction position of the component 120 by the suction nozzle 22 (center position of the component 120) from the correct suction position (center position of the suction nozzle 22) is calculated. Based on the calculated deviation amount, the head unit 20 is moved in the XY directions by the servo motors 24 and 43 and the suction nozzle 22 is rotated by the nozzle rotating device, so that the mounting position of the component 120 is corrected. The above-described correction processing of the mounting position of the component 120 is performed in parallel with the head unit 20 moving from the tape feeder 121 to a predetermined mounting position of the printed circuit board 110.

  As shown in FIG. 1, after the head unit 20 is moved to the mounting position of the printed circuit board 110, the suction nozzle 22 is lowered and the component 120 is mounted on the printed circuit board 110. By repeatedly performing the above processing, the component 120 is mounted on the printed circuit board 110.

  In addition, the printed circuit board 110 on which the mounting of the component 120 is completed is carried out from the base 1 via the pair of board conveyance conveyors 10. In this way, the mounting operation of the component 120 by the surface mounter 100 is completed. In parallel with this transport, the next printed circuit board 110 is transported toward the mounting work position.

  In the present embodiment, as described above, the image processing unit 64 is compared with the dimension data (A × B) of the image recognizable portion of the component 120, and all of the plurality of image portions 130 corresponding to the component 120 are processed. By configuring so as to detect one component circumscribing rectangle 150 that surrounds the image, the image of the component 120 including the plurality of image portions 130 can be recognized as one component (component circumscribing rectangle 150) without individually recognizing the plurality of image portions 130. ) Can be recognized. In addition, the image processing unit 64 detects one component circumscribing rectangle 150 surrounding all of the plurality of image portions 130 corresponding to the component 120 while comparing with the dimension data A × B of the component 120 set in advance. By configuring, only the dimension data A × B of the image recognizable part 120 of the component 120 to be recognized without individually registering information as feature points such as the arrangement and shape of each of the plurality of image parts 130. The component 120 can be recognized. In particular, when the shape of the part 120 is complicated, it is necessary to increase the number of feature points. On the other hand, since the dimension A × B of the part 120 already exists for each part 120, complicated setting work is required. The image of the component 120 composed of a plurality of image portions 130 can be recognized as one component (component circumscribed rectangle 150) without necessity.

  In the present embodiment, as described above, the image processing unit 64 circumscribes the plurality of image portions 130, and the X coordinate direction and the Y coordinate direction orthogonal to the X coordinate direction have the X coordinate of the component 120. By detecting a circumscribed rectangle that falls within a predetermined tolerance in comparison with the dimension A in the direction and the dimension B in the Y coordinate direction, the component circumscribed rectangle 150 that surrounds all of the plurality of image portions 130 corresponding to the component 120 is detected. With this configuration, the part circumscribing rectangle 150 can be easily detected by comparing with the dimension data A × B of the part 120 in the X coordinate direction and the Y coordinate direction, and the circumscribing that falls within a predetermined tolerance. By detecting a rectangle (such as a circumscribed rectangle 151), the component 120 can be recognized with high accuracy.

  In the present embodiment, as described above, when there are a plurality of candidate parts 154 (such as 154) for the part circumscribing rectangle 150 of the part 120, the image processing unit 64 detects a detection condition (first number) other than the dimension A × B of the part 120. The component circumscribing rectangle 150 of the component 120 is detected based on the first to fourth detection conditions), and the detection condition includes the image portion 130 (including the circumscribing rectangle (such as 154) or the circumscribing rectangle (such as 154)). By including at least one of a detection condition relating to the area of at least one of the noise 140) or a detection condition relating to the number of image portions 130 (noise 140) included in a circumscribed rectangle (such as 154), the dimension A × B of the part 120 is obtained. Even when a plurality of matching circumscribing rectangle candidates (such as 154) that match or approximate are detected, based on the first to fourth detection conditions, the part 120 outside the part is accurately detected. The tangent rectangle 150 can be detected. Further, the component circumscribing rectangle 150 of the component 120 can be detected without requiring information regarding the arrangement and position of the feature points of the component 120 and other additional information.

  Further, in the present embodiment, as described above, the first detection condition is set by summing the areas of the image portions 131 included in the circumscribed rectangle among the plurality of candidates 154 and 155 of the component circumscribed rectangle 150 of the component 120. The circumscribed rectangle 155 including the noise 141 is obtained when the circumscribed rectangle 154 having the largest value is set as the component circumscribed rectangle 150 of the component 120 so that the noise 141 exists in the vicinity of the recognition target component 120 in the captured image 201. Since the total area of the image portion 131 (noise 141) in the rectangle is smaller than the circumscribed rectangle 154 surrounding all of the image portions 131 constituting the part 120, each image portion 131 (noise) included in the circumscribed rectangle is reduced. 141) Based on the detection condition that the circumscribed rectangle 154 having the largest total area is the component circumscribed rectangle 150 of the component 120, a plurality of candidates (1 4 and 155) parts enclosing rectangle 150 that matches the dimensions A × B component 120 from can be detected (154).

  In the present embodiment, as described above, the second detection condition is included in each of the circumscribed rectangles (156 and 157) among the plurality of candidates (156 and 157) of the component circumscribed rectangle 150 of the component 120. By setting the circumscribed rectangle 156 having the largest number of image portions 132a and 132b as the component circumscribed rectangle 150 of the component 120, a small image is formed on the outer peripheral portion of the recognition target component 120 in the captured image 202 as shown in FIG. The circumscribed rectangle 156 that surrounds all of the image portions 132a and 132b constituting the component 120, such as when the portion 132b (part of the component 120) exists, is circumscribed without including the small image portion 132b (part of the component). Since the number of image portions 132a and 132b in the rectangle is smaller than the rectangle 157, each image portion 13 included in the circumscribed rectangle is displayed. Based on the detection condition that the circumscribed rectangle 156 in which the number of a and 132b is the maximum is the component circumscribed rectangle 150 of the component 120, the component circumscribed rectangle matched by the dimension A × B of the component 120 from a plurality of candidates (156 and 157) 150 (156) can be detected.

  Further, in the present embodiment, as described above, the third detection condition is the circumscribed rectangle that minimizes the area of the circumscribed rectangle among the plurality of candidates (circumscribed rectangles 158 and 159) of the component circumscribed rectangle 150 of the component 120c. 14 is a part circumscribing rectangle of the part 120c, and as shown in FIG. 14, a part approximated in size is detected by including noise 142 existing in the vicinity of the part 120c to be recognized in the captured image 203. When a plurality of circumscribed rectangle 150 candidates (circumscribed rectangles 158 and 159) are detected, the dimension A6 × B7 of the circumscribed rectangle 159 including the noise 142 becomes larger than the dimension A5 × B6 of the circumscribed rectangle 158 by the amount of the noise 142. Therefore, based on the detection condition that the circumscribed rectangle 158 having the smallest area of the circumscribed rectangle is the component circumscribed rectangle 150 of the component 120c, a plurality of candidates (158 And 159), the part circumscribed rectangle 150 (158) that matches the dimension A × B of the part 120 can be detected.

  Further, in the present embodiment, as described above, the fourth detection condition is that each image included in the circumscribed rectangle with respect to the area of the circumscribed rectangle among the plurality of circumscribed rectangles 160 and 161 approximate to the dimension of the component 120d. The circumscribed rectangle 160 having the maximum area ratio of the portion 134 (noise 143) is set as the component circumscribed rectangle 150 of the component 120d, so that the circumscribed rectangle including the noise 143 in the captured image 204 is obtained as shown in FIG. 161, since the ratio of the image portion 134 (noise 143) occupying the area of the rectangle is relatively small, the total area of each image portion 134 (noise 143) included in the circumscribed rectangle with respect to the area of the circumscribed rectangle. From a plurality of candidates (160 and 161) based on the detection condition that the circumscribed rectangle 160 having the maximum ratio is the component circumscribed rectangle 150 of the component 120d. It is possible to detect the components enclosing rectangle 150 (160) without the size 143.

  In the present embodiment, as described above, the image processing unit 64 uses at least one detection condition (first to fourth detection conditions) applied when detecting the component circumscribed rectangle 150 of the component 120 as the component 120. By being configured so as to be selectable in advance according to the type, a part 120 in which a plurality of small image portions 130 are included in the captured image 200 due to the structure, a part 120 in which the noise 140 is easily reflected due to the shape and material of the part 120, etc. It is possible to detect the component circumscribing rectangle 150 with higher accuracy by registering in advance which detection condition is to be applied when the component circumscribing rectangle 150 is detected, and selecting an appropriate detection condition according to the type of the component 120. Can do.

  In the present embodiment, as described above, the head unit 20 is provided with the nozzle rotating device (see FIG. 3) for rotating the suction nozzle 22 to correct the position and the rotational position of the suction nozzle 22 based on the image recognition result. By configuring the component 120 to be mounted in a predetermined direction at a predetermined mounting position, in addition to the mounting position of the component 120 recognized as the component circumscribed rectangle 150 based on the image recognition result, the orientation (direction) of the component ) Can also be corrected, so that it can be transferred to a predetermined mounting position in a predetermined direction with higher accuracy.

  The embodiment disclosed this time should be considered as illustrative in all points and not restrictive. The scope of the present invention is shown not by the above description of the embodiments but by the scope of claims for patent, and further includes all modifications within the meaning and scope equivalent to the scope of claims for patent.

  For example, in the above-described embodiment, an example in which the present invention is applied to the surface mounter 100 which is an example of the component transfer apparatus has been described. However, the present invention is not limited to this, and a component testing apparatus for inspecting electronic components ( The present invention may also be applied to a component transfer device other than a surface mounter such as a so-called IC handler. In addition to the component transfer device, the present invention may be applied to a single component recognition device.

  In the above embodiment, the example in which the image processing unit 64 performs component recognition based on the dimension A × B of a predetermined portion where the image of the component 120 can be recognized has been described. Is applicable not only to the external dimensions of parts but also to the case of using dimension data of an area including a plurality of parts smaller than the external dimensions.

  Moreover, although the component imaging device 50 was provided in the head unit 20 and the structure comprised so that it was movable was shown in the said embodiment, this invention is not restricted to this, You may install an imaging part fixedly. In this case, for example, the imaging unit is fixedly installed on the base 1 with the imaging direction facing upward, and the head unit 20 moves above the imaging unit while the component 120 is sucked by the suction nozzle 22. Thus, the lower surface of the component 120 can be imaged.

  In the above-described embodiment, the example in which the image processing unit 64 recognizes the component 120 of the component 120 from the captured image 200 has been described. However, the present invention is not limited thereto, and the main calculation unit of the control device 60 may be used. 61 may acquire a captured image from the image processing unit 64 and perform component recognition processing. In addition, a dedicated image recognition unit that performs only component recognition processing without controlling the imaging unit may be provided.

  Further, in the above-described embodiment, the example in which the image processing unit 64 is configured to perform the component recognition of the component 120 and the imaging control of the component imaging device 50 has been described. However, the present invention is not limited thereto, and the image processing is performed. A dedicated image recognition unit that performs only component recognition processing without performing control of the imaging unit may be provided while the unit performs only control of the imaging unit.

  Moreover, in the said embodiment, although the control apparatus 60 was shown as an example of a component recognition part, this invention is not restricted to this, The component provided with the image recognition part separately from the control apparatus which controls a surface mounter. A dedicated component recognition unit that performs only recognition may be provided.

  In the above embodiment, an example is shown in which the component circumscribing rectangle 150 is detected by applying the first to fourth detection conditions when there are a plurality of candidates for the component circumscribing rectangle 150. The present invention is not limited to this, and other detection conditions other than the first to fourth detection conditions may be set and applied according to the type of component. Any number of detection conditions may be used.

  Moreover, in this embodiment, although the example which selects the 1st-4th detection conditions applied to component recognition according to the kind of component 120 was shown, common detection conditions are applied about all the components. Also good.

  Moreover, although the detection condition applied at the time of component recognition showed the example registered beforehand in component data according to the kind of component 120, this invention is not restricted to this, Without registering beforehand, The image processing unit 64 may select the detection condition based on the shape of the component 120 or other data.

  In the above embodiment, the example in which the coordinate rotation angle θ is set to 5 degrees and the coordinate rotation angle range is set to a range of −45 degrees to +45 degrees has been described. The angle θ may be set to an angle smaller than 5 degrees or an angle larger than 5 degrees. Further, the angle range of the coordinate rotation may be set to a range other than −45 degrees to +45 degrees, such as −20 degrees to +20 degrees. The angle range of the coordinate rotation may be changed according to the angle range in which the component sucked by the suction nozzle can be tilted and sucked. At this time, the angle range of the coordinate rotation may be changed depending on the type of component.

  In the above embodiment, the difference between the minimum Y coordinate and the maximum Y coordinate is determined for the maximum Y coordinate and the minimum Y coordinate of each image portion 130 within the X coordinate range determined by the combination of the minimum X coordinate and the maximum X coordinate. Although the present invention is not limited to this, the present invention is not limited to this, and the minimum X coordinate for the maximum X coordinate and the minimum X coordinate of each image portion 130 within the Y coordinate range determined by the combination of the minimum Y coordinate and the maximum Y coordinate. And the maximum X coordinate may be obtained.

  In the above embodiment, the group of image portions 130 (noise 140) included in the obtained circumscribed rectangle is rotated to obtain an angle φ that minimizes the area of the rectangle, and the rotation angle φ that minimizes the area. In this example, it is determined whether or not the dimension of the circumscribed rectangle in FIG. 4 is equal to or close to the dimension A × B of the part 120. However, the present invention is not limited to this and is included in the circumscribed rectangle. It is not necessary to rotate the group of image portions 130 (noise 140). This is because by performing the rotation of the coordinate system, it is possible to detect even when the group is not rotated. However, when performing group rotation, it is possible to prevent the same group from being detected each time the coordinate system is rotated by registering it in advance when it can be detected as a candidate for a component circumscribed rectangle. Can do. When the rotation angle θ is set to be small, the same circumscribed rectangle may be detected many times, so that the same circumscribed rectangle (group) can be prevented from being detected many times.

20 Head unit 22 Suction nozzle 50 Component imaging device (imaging unit)
60 Control device (component recognition unit, component recognition device)
64 Image processing unit (image recognition unit)
100 surface mounter 120 component 130 image part 150 component circumscribed rectangle (circumscribed rectangle)
151, 152, 153, 153a, 154, 155, 156, 157, 158, 159, 160, 161 circumscribed rectangle 200 captured image

Claims (5)

An image recognition unit that performs image recognition based on a captured image including a plurality of image portions corresponding to the component obtained by imaging the component;
Wherein the image recognition unit, in comparison with the dimension data of a predetermined portion of the component, and detects the taken enclose the outer contact rectangle all of the plurality of image portions corresponding to the component, the circumscribed rectangle of the candidate of the component In the case where there are a plurality of parts, the part is configured to detect one circumscribed rectangle based on a detection condition other than the dimension of the part ,
Among the plurality of candidates for the circumscribed rectangle of the part, the detection condition is that the area of the circumscribed rectangle is minimized, the total area of the image portions included in the circumscribed rectangle is maximized, or the circumscribed rectangle is maximized. A plurality of requirements on whether the number of each image portion included in the rectangle is maximized, or the ratio of the total area of each image portion included in the circumscribed rectangle to the area of the circumscribed rectangle is maximized Among these, a part recognition apparatus that sets a circumscribed rectangle that satisfies at least one requirement as a circumscribed rectangle of the part .   The image recognition unit performs image recognition based on a portion corresponding to noise in addition to a plurality of image portions corresponding to the component, and also corresponds to the noise that is outside the plurality of image portions corresponding to the component. 2. The component recognition device according to claim 1, wherein the component recognition device is configured to detect one circumscribed rectangle that surrounds all of a plurality of image portions corresponding to the component without surrounding a portion to be processed. The image recognition unit circumscribes a plurality of the image parts, and the dimension in the one direction and the other direction orthogonal to the one direction is compared with the dimension data in the one direction and the other direction of the predetermined part of the component. 3. The circumscribed rectangle according to claim 1, wherein one circumscribed rectangle surrounding all of the plurality of image portions corresponding to the part is detected by detecting a circumscribed rectangle that falls within a predetermined tolerance. 4. Component recognition device . A head unit including a suction nozzle for sucking and holding the component, and for transferring the sucked component to a predetermined transfer position;
An imaging unit that images the component sucked by the suction nozzle;
A component recognition unit including an image recognition unit that performs image recognition based on a captured image including a plurality of image portions corresponding to the component imaged by the imaging unit;
Image recognition unit of the component recognition unit, while comparing the size data of a predetermined portion of the component, and detects the taken enclose the outer contact rectangle all of the plurality of image portions corresponding to the components, circumscribing the component When there are a plurality of candidate rectangles, one circumscribed rectangle of the part is detected based on a detection condition other than the dimension of the part ,
Among the plurality of candidates for the circumscribed rectangle of the part, the detection condition is that the area of the circumscribed rectangle is minimized, the total area of the image portions included in the circumscribed rectangle is maximized, or the circumscribed rectangle is maximized. A plurality of requirements on whether the number of each image portion included in the rectangle is maximized, or the ratio of the total area of each image portion included in the circumscribed rectangle to the area of the circumscribed rectangle is maximized A circumscribed rectangle that satisfies at least one of the requirements is a circumscribed rectangle of the component,
The head unit is a component transfer device configured to transfer the component to the predetermined transfer position by correcting the position of the suction nozzle based on an image recognition result. The head unit further includes a nozzle rotating device that rotationally drives the suction nozzle, and corrects the position and rotation position of the suction nozzle based on the image recognition result to place the component in a predetermined direction at the predetermined transfer position. The component transfer device according to claim 4 , wherein the component transfer device is configured to transfer to a component.

Images