JP3996568B2 - Image processing method, error missing item inspection method, image processing apparatus, and error missing item inspection apparatus - Google Patents

Description

  The present invention relates to an image processing method and an image processing apparatus for identifying an object to be imaged by performing predetermined image processing using a captured image, and an image processing method and an image processing apparatus to which the image processing method and the image processing apparatus are applied. The present invention relates to an error missing item inspection method and an error missing item inspection device for inspecting an error of an externally mounted component.

  In the assembly process of automobile parts, a process of inspecting whether or not a plurality of parts assembled to the part is correct and whether there is a missing part of the part to be assembled (hereinafter, this process is mistakenly missing) Called inspection process). For example, in the assembly process of an engine, whether or not multiple types of parts such as EGR (Exhaust Gas Recirculation) valves, oil filler caps, turbochargers, etc. assembled in the engine are correct and whether these parts are missing. It has been inspected.

  And in this mistaken missing goods inspection process, the inspection of the mistaken missing goods of an assembly part is performed using the exclusive mistaken article inspection apparatus. As described in, for example, Japanese Patent No. 3372014, the erroneous missing item inspection apparatus includes an imaging unit in which an illumination device and an imaging device are integrated at the tip of a robot arm. Is set at a predetermined image pickup position with respect to the engine, an assembly part is imaged, and a predetermined image processing is performed on the captured image to determine the presence / absence and correctness of the assembly part.

Specifically, as shown in FIGS. 10A to 10C, a captured image of an assembly part (hereinafter, this image is referred to as an original image) is subjected to edge point enhancement processing, and the edge point enhanced image is displayed. A line in the captured image (contour line of the image including the shape of the assembly part) is extracted by a well-known edge detection process, and this contour line pattern (extraction pattern) is a standard contour line prepared in advance. Compared with the pattern (standard pattern; not shown), the matching rate (%) of both patterns is calculated. For example, whether the matching rate is equal to or higher than a predetermined threshold (for example, 60%) Correctness etc. are judged.
Japanese Patent No. 3372014

  For example, the engine has a main body formed of cast iron, and has a black color as a whole. Small parts such as an EX engine bracket of a relatively dark color are assembled outside the main body. For this reason, since the entire engine becomes a blackish color, when imaging an assembly part with an erroneous part inspection device, the change in the assembly part is displayed so that a clear shade change appears in the outline part of the assembly part. An imaging unit is set at a proximity position, and the captured image is acquired by irradiating the assembly light with the illumination light.

  However, since the captured image (original image) captured by illuminating the illumination light is also applied to the engine body or the like that is the background of the assembly part, the edge of the original image is emphasized, for example, the background of the engine body (The uneven cast pattern formed on the surface of the cast iron) emerges, and the cast pattern of the engine body is extracted as a contour line in the subsequent edge detection process.

  Therefore, conventionally, when correctness / incorrectness of assembled parts is determined by pattern matching, the distribution of the coincidence rate with respect to the correct assembled parts is as shown by the solid line in FIG. 11, while the shape is similar to the correct assembled parts. The distribution of the coincidence rate with respect to the assembled parts is as shown by the dotted line in FIG. 11. The portion indicated by A is erroneously determined as an incorrect product, and the portion indicated by B is erroneously determined as an authentic product.

  As a result, if it is determined that the product is a wrong product and the assembly part is subjected to a blind inspection, for example, if the product is judged to be an erroneous judgment of the part indicated by A, there is an error in the correct assembly part. The inspection by the product inspection device and the silent inspection by the worker are performed, and there is a problem that it takes unnecessary man-hours and time.

  The present invention has been made in view of the above-described problems, and has an image processing method and an image processing apparatus with high identification accuracy and less misjudgment, and an error-missing item inspection to which the image processing method and the image processing apparatus are applied. Provided are a method and a device for inspecting an error.

  The present invention provides an image input means for inputting a digital image of an object imaged by an imaging device, a storage means for storing a standard captured image pattern corresponding to the imaged image of the object, and the digital image Density enhancement processing means for performing density enhancement of the image, a blur image generation means for generating a blurred image of the object using the digital image on which the density enhancement processing has been performed, and the object of the object using the blur image Pattern extraction means for extracting a pattern of the captured image, and comparing the pattern extracted by the pattern extraction means with a standard pattern stored in the storage means, corrects the object imaged by the imaging device. An image processing apparatus comprising: a determination unit for determining.

  The present invention also relates to an image processing method for identifying an object to be imaged by performing predetermined image processing using a captured image, and an image input step of inputting a digital image of the object imaged by the imaging device A density enhancement processing step of performing density enhancement of the digital image, a blur image generation step of generating a blur image of the object using the digital image on which the density enhancement processing has been performed, and using the blur image A pattern extraction step for extracting a pattern of a captured image of the target object, and a pattern of a standard captured image corresponding to the captured image of the target object stored in the storage means, the pattern extracted by the pattern extraction step; And a determination step of determining whether the object imaged by the imaging device is correct by comparing (claim 2).

  According to the image processing apparatus or the image processing method having the above configuration, when a digital image of an object imaged by the imaging apparatus is input, density enhancement of the digital image is performed, and the object is further detected using the digital image. The blurred image is generated. A pattern of a captured image of the object is extracted using the blurred image, and an image is captured by comparing the extracted pattern with a pattern of a standard captured image stored in the storage unit (hereinafter referred to as a standard pattern). The correctness / incorrectness of the selected object is determined.

  According to the present invention, since the pattern of the captured image of the object is extracted using the blurred image, a portion having a density change in the shape of the object or the background or the periphery of the object remains as a contour line by the density enhancement process. Even in this case, a portion where there is no density difference equal to or greater than a certain value is unclear in the blurred image and is not extracted as a pattern of the captured image. Therefore, since only a portion having a relatively large density change such as a contour portion indicating the shape of the target object is extracted as a pattern constituting the captured image, the accuracy of the target correct / incorrect determination by comparison with the standard pattern is improved. To do.

  That is, for example, when the correctness of an object is determined based on whether the matching rate between the extracted pattern and the standard pattern exceeds a predetermined threshold, the distribution of the matching rate when the target is correct is relatively higher than the predetermined threshold. On the other hand, the distribution of coincidence rate when the object is wrong gathers in the part of the value relatively lower than the predetermined threshold value, and the base of both distributions does not spread near the threshold value. The accuracy of correct / incorrect determination is improved.

  The present invention includes an image pickup unit that picks up an image of an externally mounted part of a workpiece having at least one externally mounted part, a position setting unit that sets the image pickup unit at a predetermined image pickup position of the work, and the externally mounted part. Storage means for storing a standard captured image pattern corresponding to the captured image, density enhancement processing means for enhancing the density of the digital image of the externally mounted part imaged by the imaging means, and the density enhancement processing A blurred image generating unit that generates a blurred image of the externally mounted component using the digital image, a pattern extracting unit that extracts a pattern of a captured image of the externally mounted component using the blurred image, and the above An external image captured by the imaging device by comparing the pattern extracted by the pattern extracting means with a standard pattern stored in the storage means And determining component determining means correctness of Chakubuhin a Ayamaketsu article inspection apparatus is characterized in that and an output means for outputting visible to the determination result of the component determination means (claim 3).

  According to the present invention, the externally mounted component of the workpiece having at least one externally mounted component is imaged by an imaging device, and the component externally mounted on the workpiece using the captured image output from the imaging device. An error-missing-item inspection method for inspecting correctness / incorrectness, wherein the imaging device is set at a predetermined imaging position of the workpiece to image the externally mounted component, and the externally mounted component imaged by the imaging device A density enhancement processing step for enhancing the density of the digital image, a blur image generation step for generating a blur image of the externally mounted component using the digital image subjected to the density enhancement processing, and using the blur image A pattern extracting step for extracting a pattern of a captured image of the externally mounted component, and the externally mounted component in which a pattern extracted by the pattern extracting step is stored in a storage unit Comparing with a standard captured image pattern corresponding to a captured image, a component determination step for determining the correctness of an externally mounted component imaged by the imaging device, and a determination result of the component determination step are output visually. And an output step (claim 4).

  According to the error missing item inspection apparatus or the error missing item inspection method having the above-described configuration, when a digital image of the externally mounted part of the workpiece imaged by the imaging device is input, density enhancement of the digital image is performed. A blurred image of the object is generated using the digital image. Then, the pattern of the captured image of the externally mounted component is extracted using the blurred image, and the correctness of the externally mounted component captured by comparing the extracted pattern with the standard pattern of the externally mounted component stored in the storage unit is determined. Determined.

  According to the present invention, since the pattern of the captured image of the externally mounted component is extracted using the blurred image, a portion having a density change in the shape of the externally mounted component or the background of the object or the periphery thereof by the density enhancement process is a contour line. If there is no density difference equal to or greater than a certain value even if it remains, the blurred image is unclear and is not extracted as a pattern of the captured image. For example, in the inspection of externally mounted parts of an engine, if dark parts are mounted on an engine formed of black cast iron and density enhancement processing is performed on the captured image of the parts, not only the contour part of the shape of the part The density gradient is also enhanced for the convex and concave portions of the engine background, but since the pattern is extracted after the blurred image is generated, the convex and concave portions of the engine background are unclear and are not extracted as a pattern.

  Accordingly, only a portion having a relatively large density change, such as a contour portion indicating the shape of the externally mounted component, is extracted as a pattern constituting the captured image. For example, the reason described above (correction in the determination based on the pattern matching rate) The accuracy of the correct / incorrect determination of the object by comparison with the standard pattern is improved from the reason that the difference in distribution between the parts and the wrong parts becomes clear.

  According to the present invention, since the pattern of the captured image of the shooting target is extracted using the blurred image, it is possible to improve the accuracy of the correctness determination of the shooting target by pattern matching between the extracted pattern and the standard pattern. In particular, when the present invention is applied to an apparatus that determines the correctness / incorrectness of a component mounted on the outside of a workpiece from a captured image, it is possible to determine whether the externally mounted component is correct / correct with high accuracy.

  An error missing item inspection method and an error missing item inspection apparatus according to the present invention will be described with reference to the drawings.

  FIG. 1 is a diagram showing a schematic configuration of an embodiment of an error missing item inspection apparatus according to the present invention, and FIG. 2 is a front view of an imaging unit provided at the arm tip of the error missing item inspection apparatus.

  The error missing item inspection apparatus 1 includes a robot 2 composed of a vertical articulated robot, an imaging unit 3 provided at the tip of an arm 21 of the robot 2, and a control device 4 that controls operations of the robot 2 and the imaging unit 3. And an input / output display device 5 for inputting information necessary for control to the control device 4 and displaying a control result.

  As shown in FIG. 2, the imaging unit 3 has a horizontally-long rectangular box shape when viewed from the front, and the approximate center of the bottom surface is fixed to the arm 21 of the robot 2. In addition, two illumination light projecting units 32a and 32b and a color imaging unit 31a are alternately arranged in a row along the long side at the upper part of the front, and three along the long side at the lower part of the front. Illuminating light projecting sections 32c, 32d, 32e and monochrome imaging sections 31b, 31c are alternately arranged in a line.

  The color imaging unit 31a includes a color area sensor made up of, for example, a CCD (Charge-Coupled Device) and a zoom lens that forms an optical image of a subject on the imaging surface of the color area sensor. The color area sensor is composed of a single-plate color area sensor in which R (red), G (green), and B (blue) color filters are arranged in a checkered pattern on the imaging surface of each pixel. The color imaging unit 31a is for performing color imaging of the assembly component 7 in order to identify the assembly component 7 by color. That is, depending on the assembly part 7, the appearance shape may be exactly the same and only the performance may be different. In such a case, the color mark is attached to the assembly part 7, so that the color mark of the assembly part 7 is determined from the captured image. The type of the assembly component 7 is identified by detecting the color.

  The monochrome imaging units 31b and 31c are also composed of, for example, a monochrome area sensor composed of a CCD (Charge-Coupled Device) and a zoom lens that forms an optical image of a subject on the imaging surface of the color area sensor. The reason why two monochrome imaging units are provided is that these monochrome imaging units 31b and 31c constitute a stereo camera so that a three-dimensional image of the subject can be acquired. The color area sensor and the monochrome area sensor are not limited to the CCD, but may be other solid-state imaging devices such as a C-MOS image sensor. Further, the color area sensor is not limited to a single-plate type, and may be a three-plate type color image sensor provided with three image sensors corresponding to each color of R, G, and B.

  The illumination light projecting units 32a to 32e are configured by a bundle of a number of optical fibers, guide light emitted from a light source (not shown) to the front surface (imaging surface) of the imaging unit 3, and the color imaging unit 31a and the monochrome imaging unit. Irradiate toward the subjects 31b and 31c.

  The imaging unit 3 can be moved to a predetermined imaging position with respect to the engine 6 by driving the arm 21 of the robot 2, and at the imaging position, the imaging unit 3 is arbitrary in the XY plane (a plane parallel to the front of the imaging unit 3). It can be rotated. Accordingly, the conditions of the shooting position and shooting direction of the imaging unit 3 with respect to each assembled component 7 of the engine 6 are set in advance by controlling the drive of the arm 21 of the robot 2 and the magnification of the zoom lens. (An angle-of-view condition necessary for image matching processing in the erroneous missing item inspection unit 42) is set.

  The control device 4 includes a microcomputer having a basic configuration of a CPU (Central Processing Unit), a ROM (Read Only Memory), a RAM (Random Access Memory), and an input / output IF (Interface), and a program stored in advance in the ROM is stored in the CPU. For example, using the function of controlling the operation of the robot 2 and the imaging unit 3 and the image captured by the imaging unit 3, for example, whether or not the component 7 assembled outside the engine 6 is correct or not is determined. It performs the functions of image processing. The robot control unit 41 in the control device 4 is a functional block showing a part that controls the operation of the robot 2 and the imaging unit 3, and the erroneous part inspection unit 42 is an assembly part 7 that uses the captured image. The part which performs the determination process of the right / wrong and the presence / absence of a missing part is shown by a functional block.

 The ROM has a control procedure for the robot 2 for imaging a plurality of assembly parts 7 to be inspected for each type of engine 6 by the imaging unit 3 and an imaging procedure for the imaging unit 3 (a plurality of assembly parts 7 are provided). In this case, a program of a procedure for imaging each assembly part 7 in order is stored, and the robot control unit 41 develops this program in the RAM and executes it to execute the assembly part 7 in the erroneous part inspection process. The capture of the captured image is controlled. In other words, in the present embodiment, the misplaced product inspection unit 42 uses the EGR valve, the outlet water, the IN insulator, the oil filler cap, the alternator, the turbocharger, the EX engine bracket, and the EX insulator 7 types of assembly parts 7 of the engine 6. Whether the parts are correct or not and whether there is a missing part is determined, and the ROM stores information such as the imaging position of the assembled part 7 and the imaging direction of the imaging unit 3 for each of the assembled parts 7. The robot control unit 41 drives the arm 21 of the robot 2 according to this information to set the imaging unit 3 at a predetermined imaging position with respect to the assembly component 7 to be imaged, and the imaging unit 31 ( The captured image of the assembly component 7 is captured by driving (described later). The captured image data is input to the error / missing part inspection unit 42 and is used for processing (described later) for determining whether the assembled part 7 is correct or not and whether there is a missing part.

  The ROM stores a standard pattern (hereinafter referred to as a standard pattern) used when the error / missing item inspection unit 42 determines the correctness of the assembled part 7 and the presence / absence of a missing item by pattern matching. Yes. The required number of standard patterns is prepared for each of the eight types of assembly parts 7 of the EGR valve or EX insulator described above (if there are multiple types of parts for each assembly part, the number of types). Yes.

  Each standard pattern includes a contour line (a contour line indicating the shape of the assembly component 7 and a typical contour line that appears in the background) included in the captured image when the corresponding assembly component 7 is captured at a predetermined position. Consisting of). For example, when there are five types of outlet water having different shapes for the outlet water of the engine 6, five types of standard patterns are stored in the ROM as shown in FIGS. In FIGS. 3A to 3E, a portion surrounded by a square frame is an imaging range (standard pattern), and each standard pattern includes an outline of an outlet water shape included in the frame and a background portion. The representative contour line (for example, the contour line of the cylindrical main body portion to which the cap is attached).

  The input / output display device 5 is composed of a touch panel, and the input / output display device 5 is inputted with conditions necessary for the inspection of the missing item (for example, the type and name of the engine to be inspected and the type and name of the assembly parts of the engine) A menu screen is displayed, and a required condition is input to the control device 4 when the operator presses an operation switch or a selection item displayed on the menu screen.

  Next, the function of the erroneous product inspection unit 42 according to the present invention will be specifically described with reference to FIG.

  FIG. 4 is a functional block diagram showing the processing steps of the erroneous product inspection unit 42. The erroneous missing item inspection unit 42 includes an image enhancement processing unit 421, a blurred image creation unit 422, a contour line extraction unit 423, a pattern matching processing unit 424, and a component determination unit 425. The standard pattern storage unit 426 stores the standard pattern of the assembly component 7 used for pattern matching. As described above, the ROM performs its function as described above. Note that an EEPROM may be provided in addition to a ROM that stores a program for performing drive control of the robot 2 and the imaging unit 3 and an erroneous missing item inspection process, and a standard pattern may be stored in the EEPROM.

The image enhancement processing unit 421 performs image enhancement processing on digital image data (hereinafter referred to as original image data) obtained by imaging the assembly component 7 by the monochrome imaging units 31b and 31c of the imaging unit 2. . The image enhancement processing unit 421 generates sharpened original image data (hereinafter referred to as “emphasized image data”) by superimposing image data obtained by edge-enhancing the original image on the original image data. Specifically, for example, the original image data is composed of density data of a pixel group of x columns and y rows, a function representing the original image is f (x, y), and a secondary of the original image function f (x, y). When the differential function is ∇ ² f (where Laplacian ∇ ² is a second-order differential operator and ∇ ² = ∂ ² / ∂x ² + ∂ ² / ∂y ² ), the image enhancement processing unit 421 , G (x, y) = f (x, y) −∇ ² f (x, y) is calculated to calculate a function g (x, y) representing the enhanced image.

  The image enhancement processing in the image enhancement processing unit 421 is the same as the image enhancement processing in the conventional inspection method described above, and as shown in FIG. 10, the edge enhancement is performed from the original image (see FIG. 10A) by this image enhancement processing. Is obtained (see FIG. 10B).

  The blurred image creation unit 422 creates a blurred image using the enhanced image data output from the image enhancement processing unit 421. The blurred image creation unit 422 generates a blurred image by performing an averaging process or an integral operation on the emphasized image. Specifically, the blurred image creation unit 422 generates a blurred image by selectively averaging density data of local portions using, for example, a 3 × 3 mask. That is, if a function representing a blurred image is h (x, y), h (x, y) = {g (x−1, y−1) + g (x, y−1) + g (x + 1, y−1). ) + G (x-1, y) + g (x, y) + g (x + 1, y) + g (x-1, y + 1) + g (x, y + 1) + g (x + 1, y + 1)} / 9 An image is created. By this calculation processing, for example, in the example of FIG. 10B, a blurred image as shown in FIG. 5 is generated.

  The contour line extraction unit 423 generates a contour line (contour line indicating the shape of the assembled part 7 or the background part to which the assembled part 7 is attached) from the blurred image created by the blurred image creation part 422. Extract outlines of other parts that appear). The contour line extraction unit 423 extracts a contour line by performing a differentiation operation on pixel data constituting a blurred image using, for example, a second-order differential filter (Laplacian filter or Gaussian Laplacian filter). By this contour line extraction process, for example, in the example of FIG. 5, a contour line pattern is obtained as shown in FIG.

  The pattern matching processing unit 424 collates the contour pattern extracted by the contour extraction unit 423 (hereinafter referred to as “extraction pattern”) with the standard pattern of the assembly component 7 stored in the standard pattern storage unit 426, and The pattern matching rate is calculated. The pattern matching processing unit 424 calculates, for example, a positional deviation amount (a distance between corresponding points) between the standard pattern and the extracted pattern, and calculates a ratio of patterns whose positional deviation amount is a predetermined threshold value or less. For example, when all the contour lines included in the extracted pattern are within a range of deviation amounts below a predetermined threshold with respect to the contour line of the standard pattern, the matching rate R is determined to be 100%. On the other hand, when all the contour lines included in the extracted pattern are in a range of an amount larger than a predetermined threshold with respect to the contour line of the standard pattern, the matching rate R is determined to be 0%. Further, among all the contour lines included in the extracted pattern, for example, when the half contour line is within the range of the deviation amount below the predetermined threshold with respect to the contour line of the standard pattern, the matching rate R is 50%. It is determined.

  As described above, in the present embodiment, the image enhancement processing is performed on the captured image of the assembly component 7, and then the blurred image is generated and the pattern is extracted from the blurred image. Extraction of a cast surface portion of a cast product other than a part as a pattern is reduced, and the distribution of the coincidence rate corresponding to FIG. 11 is as shown in FIG. 7, for example. Therefore, as shown in the figure, the distribution of coincidence rate for correct assembly parts (distribution indicated by solid lines) and the distribution of coincidence rate for incorrect assembly parts similar in shape to correct assembly parts (indicated by dotted lines) (Distribution) does not overlap in the vicinity of the threshold value Rt as in the conventional case (see FIG. 11), so that this overlapping portion is not erroneously determined, and the inspection accuracy is improved as compared with the conventional case. .

  As described above, the standard pattern storage unit 426 stores a plurality of types of standard patterns when there are a plurality of types of parts having different shapes for each assembled part 7. Therefore, when there are a plurality of types of standard patterns for a certain assembled part 7, the pattern matching processing unit 424 determines not only the standard pattern corresponding to the ID number of the assembled part 7 but also other standard patterns. The coincidence rate R is calculated. That is, as shown in Table 1, for example, there are five types of assembly parts 7: a, b, c, d, and e. For each assembly part 7, five types of A, B, C, D, and E For example, for the assembly part a, the pattern matching rate R is calculated for each of the five standard patterns A, B, C, D, and E.

  As described above, the reason why the coincidence rate R is calculated for the standard patterns other than the standard pattern corresponding to the ID number and the determination of an erroneous product is made is to increase the reliability of the determination result. That is, when it is determined whether or not the assembled part 7 is correct or missing from the pattern matching rate R only for the standard pattern corresponding to the ID number, for example, the incorrect part is determined to be a correct part due to noise included in the captured image. If the correct part is determined to be a wrong product, it cannot be confirmed, and an erroneous determination will occur. However, the standard pattern other than the standard pattern corresponding to the ID number is Since it is confirmed even if there is a possibility that the above-mentioned misjudgment may occur when judging whether the assembled parts 7 are correct or not from the coincidence rate R, the occurrence of misjudgment should be prevented as much as possible. Because you can.

  The component determination unit 425 determines whether the assembled component 7 is correct or not based on the pattern matching rate R calculated by the pattern matching processing unit 424. The component determination unit 425 compares the pattern matching rate R with a predetermined first threshold value Rt1 (for example, 60%), and determines that the component is a correct component if the matching rate R is equal to or greater than the first threshold value Rt1. When the coincidence rate R is smaller than the first threshold value Rt1, it is determined that the product is a wrong product. Further, the component determination unit 425 compares the pattern matching rate R with a predetermined second threshold Rt2 (<Rt, for example, 10%) set in advance, and if the matching rate R is equal to or higher than the second threshold Rt2, the component If it is determined that there is a match and the coincidence rate R is smaller than the second threshold value Rt2, it is determined that the product is missing.

  As described above, the component determination unit 425 performs determination processing for a plurality of standard patterns for each assembled component 7. That is, in the example of Table 1, for example, the matching rate R and the first threshold value Rt1 and the second threshold value Rt2 are set for five standard patterns A, B, C, D, and E for the assembly part 7 of a. By comparison, the presence / absence and correctness of parts are determined for each standard pattern. For example, when the determination result is as shown in Table 2, since R> Rt1 only for the standard pattern A, the inspection result of the assembled part 7 of “a” is “OK”. On the other hand, when the determination result is as shown in Table 3, for example, since R> Rt1 for the standard patterns A and D, the inspection result of the assembled part 7 of “a” is “NG”. That is, in the case of Table 3, since the assembly part 7 cannot be discriminated from the part A or the part D, in order to increase the inspection accuracy even if the standard pattern corresponding to the ID number is the part A, It will be sent to the silent inspection as "NG" treatment.

  In Table 3, when only the standard pattern of D satisfies R> Rt1, the inspection result is “NG”. In this case, since R> Rt1 with respect to the standard pattern of the D-type component, it is recognized that the D-type component 7 is attached where the A-type component 7 should be originally attached. It is.

  Next, the processing operation of the erroneous product inspection unit 42 according to the present invention will be described with reference to FIGS.

  FIG. 8 is a flowchart showing the processing procedure of the error missing part inspection unit 42.

  When a workpiece to be inspected, for example, the engine 6 is set at a predetermined inspection position (S1: YES), the ID number of the engine 6 is read (S2). The ID number of the engine 6 is read by imaging the ID number described at a predetermined position of the engine 6 by the imaging unit 3 and analyzing the captured image. When the ID number of the engine 6 is confirmed, one or two or more assembly parts 7 to be inspected by the engine 6 are specified, and the assembly parts 7 are set in the order preset by the imaging unit 3. The captured image is used to check the correctness of parts and the presence / absence of missing parts for each assembled part 7.

  That is, first, 1 is set to a counter n that counts the number of assembly parts 7 to be inspected (S3). Then, from ROM, No. The image pickup conditions (conditions such as illumination amount, image pickup position, and image pickup direction) for the assembly part 7 of the first assembly are read out, and the robot 2 is driven based on the image pickup conditions and the image pickup unit 3 is driven to A captured image for one assembly component 7 is captured (S4). In this imaging, color imaging or monochrome imaging, or both color imaging and monochrome imaging are performed depending on the type of the assembly component 7, but in this description, the case of monochrome imaging will be described for convenience.

  Subsequently, using the captured image (original image) captured by the imaging unit 3 in steps S5 to S9, No. 4 is used. The determination process of the presence / absence of one assembly part 7 and the presence or absence of a missing part is performed. That is, first, after the original image enhancement processing is performed by the image enhancement processing unit 421 (S5), a blurred image is generated by the blurred image generation unit 422 using this enhanced image (S6), and further the contour line An outline (pattern) included in the captured image is extracted from the blurred image by the extraction unit 423 (S7).

  Subsequently, the pattern matching processing unit 424 performs No. The match rate R of the extracted pattern with respect to one or more standard patterns provided for one assembly component 7 is calculated (S8), and thereafter, the component determination unit 425 sets the first threshold value Rt1 and the first threshold Rt1 for each match rate R. 2 by determining the magnitude relationship with the threshold value Rt2. It is determined whether or not there is a component error or missing item for one assembled component 7 (S9). This determination result is temporarily stored in the RAM (S10).

  Subsequently, after the counter n is incremented by 1 (S11), it is determined whether or not the counter value n exceeds the inspection number N of the engine 6 (S12). The inspection number N is set corresponding to the ID number, and can be known by reading the ID number in step S2. If n ≦ N in step S12, No. The process returns to step S4 to perform the determination process for the n assembled parts 7. And it is No. by the method similar to the determination method mentioned above. n, the determination process is performed on the assembled parts 7, and the same method is used for No. 5 below. When the determination process is completed up to N assembly parts 7, since n> N in step S12, the process proceeds to step S13, and the determination results of the N assembly parts 7 are obtained as the inspection results of the engine 6, for example, FIG. After being displayed on the input display device 5 in the display mode shown in (1), the work is carried out (S14), and the inspection process ends.

  In addition, the display mode shown in FIG. 9 is provided with “OK”, “NG” and “inspection” display windows for all the assembly parts 7 to be judged, and corresponds to the judgment results of the respective assembly parts 7. The display window to be turned on is turned on. “Total OK” and “Total NG” display indicate the overall inspection result, and “ID reading OK” and “ID reading NG” indicate whether the ID reading process of the engine 6 in the inspection process is correct or not. To display. When “Comprehensive NG” is turned on, “NG” is turned on in any of the assembled parts 7 or “ID reading NG” is turned on so that the cause of NG can be understood. .

  As described above, the erroneous missing item inspection apparatus 1 generates a blurred image from the captured image of the assembly component 7 and extracts a pattern from the blurred image. It is possible to reduce the accuracy, and it is possible to improve the accuracy of determination as to whether the assembled part 7 is correct or not and whether there is a missing part. In particular, it is possible to appropriately prevent the background portion of the casting from being mixed as noise during pattern extraction for the determination processing of the component 7 assembled to the cast iron component 6 such as an engine, and the assembly component 7 with high accuracy. Can be determined.

It is a figure which shows schematic structure of one Embodiment of the mistaken item inspection apparatus which concerns on this invention. It is a front view which shows the imaging unit provided in the arm front-end | tip of the error | missing item inspection apparatus which concerns on this invention. It is a figure which shows an example of five types of standard patterns of outlet water. It is the figure which represented the process process of the mistake missing item inspection part with the functional block. It is a figure which shows an example of the blur image produced from the image of FIG.10 (b). It is a figure which shows an example of the pattern from which a blur image is extracted. It is a figure which shows an example of distribution of the coincidence rate of the extraction pattern in the inspection method which concerns on this invention. It is a flowchart which shows the process sequence of a mistake missing item inspection part. It is a figure which shows an example of the display mode of a test result. It is a figure for demonstrating the procedure which extracts a pattern from the picked-up image of assembly | attachment components. It is a figure which shows an example of distribution of the matching rate of the extraction pattern in the conventional inspection method.

Explanation of symbols

DESCRIPTION OF SYMBOLS 1 Missing item inspection apparatus 2 Robot 3 Imaging unit 31 Imaging part (imaging means)
31a Color imaging unit 31b, 31c Monochrome imaging unit 32 Illumination unit 32a to 32e Light projection unit 4 Controller 41 Robot control unit (position setting means)
42 Missing product inspection section 421 Image enhancement section (density enhancement means)
422 Blur image creation unit (blur image generation means)
423 Outline extraction unit (pattern extraction means)
424 pattern matching processing unit 425 component determination unit (determination unit, component determination unit)
426 Standard pattern storage unit (storage means)
5 Input / output display device (output means)
6 Engine (work)
7 Parts (Externally mounted parts)

Claims (4)

Image input means for inputting a digital image of the object imaged by the imaging device;
Storage means for storing a standard captured image pattern corresponding to the captured image of the object;
Density enhancement processing means for performing density enhancement of the digital image;
A blurred image generating means for generating a blurred image of the object using the digital image on which the density enhancement processing has been performed;
Pattern extraction means for extracting a pattern of a captured image of the object using the blurred image;
Determining means for comparing the pattern extracted by the pattern extracting means with a standard pattern stored in the storage means to determine whether the object imaged by the imaging device is correct;
An image processing apparatus comprising: An image processing method for identifying an object to be imaged by performing predetermined image processing using a captured image,
An image input step of inputting a digital image of the object imaged by the imaging device;
A density enhancement processing step for density enhancement of the digital image;
A blurred image generation step of generating a blurred image of the object using the digital image subjected to the density enhancement processing;
A pattern extraction step of extracting a pattern of a captured image of the object using the blurred image;
By comparing the pattern extracted in the pattern extraction step with a pattern of a standard captured image corresponding to the captured image of the object stored in the storage means, the correctness of the object imaged by the imaging device is determined. A determination step to
An image processing method comprising: Imaging means for imaging the externally mounted component of the workpiece having at least one externally mounted component;
Position setting means for setting the imaging means to a predetermined imaging position of the workpiece;
Storage means for storing a standard captured image pattern corresponding to the captured image of the externally mounted component;
Density enhancement processing means for performing density enhancement of a digital image of the externally mounted component imaged by the imaging means;
A blurred image generating means for generating a blurred image of the externally mounted component using the digital image subjected to the density enhancement process;
Pattern extraction means for extracting a pattern of a captured image of the externally mounted component using the blurred image;
Component determining means for comparing the pattern extracted by the pattern extracting means with a standard pattern stored in the storage means to determine the correctness of the externally mounted component imaged by the imaging device;
An output means for outputting the determination result of the component determination means in a visible manner;
A device for inspecting an erroneous product, comprising: An error in which an externally mounted component of a workpiece having at least one externally mounted component is imaged by an imaging device and the correctness of the component externally mounted on the workpiece is inspected using a captured image output from the imaging device Product inspection method,
An imaging step of setting the imaging device to a predetermined imaging position of the workpiece and imaging the externally mounted component;
A density enhancement processing step for performing density enhancement of the digital image of the externally mounted component imaged by the imaging device;
A blurred image generation step of generating a blurred image of the externally mounted component using the digital image on which the density enhancement processing has been performed;
A pattern extraction step of extracting a pattern of a captured image of the externally mounted component using the blurred image;
By comparing the pattern extracted in the pattern extraction step with a pattern of a standard captured image corresponding to the captured image of the externally mounted component stored in the storage unit, the correctness of the externally mounted component captured by the imaging device is compared. A component determination process for determining
An output process for visually outputting the determination result of the component determination process;
A method for inspecting an erroneous product, characterized by comprising: