JP7664865B2 - Processing support device and processing system - Google Patents

Description

The present invention relates to a processing support device and a processing system.

Patent document 1 shows a device for detecting surface defects on a workpiece after polishing in a polishing machine, and for re-polishing the workpiece so that the detected surface defects are removed.

JP 2016-78150 A

The workpiece surface may contain a variety of surface patterns, from surface patterns that should be removed by processing to surface patterns that remain after processing. In such cases, it is not possible to easily detect areas that need to be reprocessed, as in Patent Document 1.

The present invention aims to provide a processing support device and processing device that can properly detect areas that require reprocessing.

The processing support device according to the present invention comprises:
A presentation unit that presents information on a plurality of types of work surfaces;
A designation unit capable of externally designating one of the multiple types of work surface information presented by the presentation unit;
An imaging unit for acquiring an image of a workpiece to be processed after the workpiece is processed;
a re-machining determination unit that determines a portion of the workpiece that needs to be re-machined based on information designated by the designation unit and information obtained from the captured image;
Equipped with
The information designated by the designation unit is designated as information indicating a surface condition requiring reprocessing .
(1)
The processing support device according to one aspect of the present invention comprises:
Furthermore, the information of the workpiece surface presented by the presentation unit is information indicating the spatial frequency components of the workpiece surface.
(2)
In addition, a processing support device according to another aspect of the present invention is
Furthermore, the information obtained from the captured image is information indicating the spatial frequency components of the workpiece surface in the captured image.
(3)
In addition, a processing support device according to another aspect of the present invention is
Furthermore, the photographing unit acquires photographed images of a plurality of locations of the workpiece to be processed,
The image processing device further includes an AnoGAN processing unit that divides the captured images of the plurality of locations into a first image group and a second image group having a larger number of areas detected as abnormal than the first image group by AnoGAN,
The re-processing determination unit does not determine the locations of the captured images in the first image group that require re-processing, but determines the locations of the captured images in the second image group that require re-processing.

A processing system according to another aspect of the present invention includes:
An imaging unit for acquiring an image of a workpiece to be processed after the workpiece is processed;
a re-machining determination unit that determines a portion of the workpiece that requires re-machining based on information indicating a surface state that requires re-machining and information obtained from the captured image;
Equipped with.
(1)
According to another aspect of the present invention, there is provided a processing support device,
Furthermore, the information obtained from the captured image is information indicating the spatial frequency components of the workpiece surface in the captured image.
(2)
In addition, a processing support device according to another aspect of the present invention is
Furthermore, the photographing unit acquires photographed images of a plurality of locations of the workpiece to be processed,
The image processing device further includes an AnoGAN processing unit that divides the captured images of the plurality of locations into a first image group and a second image group having a larger number of areas detected as abnormal than the first image group by AnoGAN,
The re-processing determination unit does not determine the locations of the captured images in the first image group that require re-processing, but determines the locations of the captured images in the second image group that require re-processing.

The processing system according to the present invention comprises:
A processing device that performs processing;
The above processing support device,
Equipped with.

The present invention provides a processing support device and a processing device that can properly detect areas that require reprocessing.

1 is a block diagram showing a processing system according to an embodiment of the present invention. FIG. 2 is a diagram for explaining an example of image conversion processing executed by an image processing unit, showing an image before conversion (A) and an image after conversion (B). 3A to 3C are diagrams showing examples of a workpiece surface image, an FFT processed image, and an IFFT image processed by an image processing unit. 1A is a diagram illustrating a learning process and a judgment process executed by an AnoGAN processing unit. FIG. 13 is a flowchart showing a surface condition designation process executed by the support processing unit. FIG. 13 is a display image diagram showing an example of a designation screen for designating a surface condition requiring reprocessing. 11 is a flowchart showing an automatic machining process executed by a machining control unit and a re-machining determination process.

Hereinafter, an embodiment of the present invention will be described in detail with reference to the drawings. FIG. 1 is a block diagram showing a machining system according to an embodiment of the present invention. The machining system 1 according to this embodiment is a system that can determine areas that require remachining after machining a workpiece K to be machined, and remachine those areas. Such processing is performed automatically or semi-automatically. Semi-automatic means that some operations required for processing (for example, operations to start each stage: an operation to start machining, an operation to start determining areas that require remachining, an operation to start remachining, etc.) are performed by the user, and the rest are performed automatically.

Furthermore, the machining system 1 according to this embodiment has a function of presenting information on multiple types of surfaces of the workpiece Ka to the user in advance, and allowing the user to specify those that require re-machining from the multiple types of surface information presented. The surface information specified by the user determines the locations that require re-machining in the machining process of the workpiece K to be machined. Hereinafter, the process of having the user specify surface information that requires re-machining will be referred to as the "surface condition specification process."

In this embodiment, the workpiece used to present multiple types of surface information to the user during the surface condition specification process and the workpiece that is actually machined are distinguished and referred to as the workpiece Ka and the workpiece to be machined K, respectively. The workpiece Ka and the workpiece to be machined K may be different, and for example, when machining multiple workpieces K to be machined, the workpiece K to be machined first may be referred to as the workpiece Ka used to present multiple types of surface information to the user.

The processing system 1 includes a processing device 11, an image capturing unit 12 that captures an image of the surface of the workpiece K to be processed, a setting device 15 used in the process of specifying the surface condition, and a control device 20 that drives and controls the processing device 11 and controls the process of specifying the surface condition. The setting device 15 includes a display (corresponding to the "presentation unit" according to the present invention) 15A that outputs information on the multiple surfaces of the workpiece Ka to the user, and an input device (corresponding to the "designation unit" according to the present invention) 15B that allows the user to perform a designation operation.

The processing device 11 is, for example, a polishing device that polishes the surface of the workpiece K to be processed. The processing device 11 includes a head unit 11a that performs the polishing and a robot arm 11b that moves the head unit 11a, and can move the head unit 11a to various locations on the surface of the workpiece K to be processed to polish those locations.

The photographing unit 12 is a digital camera or the like, and photographs the processing surface of the workpiece K to be processed. The photographing unit 12 may be configured to photograph the workpiece K to be processed so that the entire area of the processing surface of the workpiece K to be processed fits within one image, as long as it is capable of photographing still images. Alternatively, the photographing unit 12 may be configured to photograph the workpiece K to be processed so that a portion of the processing surface of the workpiece K to be processed fits within one image, and may be configured to be capable of changing the photographing position (for example, by being attached to the robot arm 11b), and may be configured to photograph the entire area of the processing surface of the workpiece K to be processed by photographing the workpiece K to be processed from various photographing positions.

The display device 15A is a device capable of outputting images. The input device 15B is a device, such as a mouse or keyboard, that allows the user to give instructions and operations to the image displayed and output on the display device 15A.

The control device 20 is a computer that has a CPU (Central Processing Unit), a RAM (Random Access Memory) in which the CPU expands data, a storage device that stores the control program and control data executed by the CPU, and an interface that transmits and receives data between the CPU and external devices. In the control device 20, multiple functional modules are realized by the CPU executing the control program.

The multiple functional modules include a support processing unit 21 that executes a surface condition designation process (a process that allows the user to designate surface information that requires reprocessing), an image processing unit 22 that performs image processing on the captured image, a processing control unit 23 that controls the drive of the processing device 11, a reprocessing determination unit 24 that determines areas that require reprocessing when processing the workpiece K to be processed, and an AnoGAN processing unit 25 that assists in the process of determining areas that require reprocessing. The image processing unit 22 functions as the first image processing unit and the second image processing unit according to the present invention.

<Unit image>
The image processing unit 22, the re-machining determination unit, and the AnoGAN processing unit 25 of the control device 20 process images (hereinafter referred to as "unit images") obtained by cutting the captured image of the workpiece K to be machined into a predetermined size. The size of the unit image may be set to correspond to the size of a section for inspecting the surface condition of the workpiece K to be machined and determining whether or not to re-machine the workpiece K.

<Image Processing Unit>
The image processing unit 22 performs coordinate conversion processing and spatial frequency component extraction processing. Fig. 2 is a diagram for explaining an example of image conversion processing executed by the image processing unit, showing an image before conversion (A) and an image after conversion (B). Fig. 3 is a diagram showing an example of a work surface image, an FFT processed image, and an IFFT image processed by the image processing unit.

The coordinate conversion process is a process of converting the coordinates of multiple unit images showing the machining surface of the workpiece K to be machined, photographed by the photographing unit 12, so that the machining marks that appear as curves become straight. In the coordinate conversion process, the image processing unit 22 first performs image recognition to extract the machining mark line L1 contained in the unit image D1, as shown in FIG. 2(A), for example, and then obtains a coordinate conversion equation that makes the line L1 a straight line. The image processing unit 22 then applies the above coordinate conversion to the original unit image D1, as shown in FIG. 2(B), to obtain a unit image D2 in which the machining mark line L1 has been converted into a straight line. It is preferable that the above coordinate conversion equation be determined with the constraint that the difference between the spacing between the machining marks in the image and the actual spacing between the machining marks is not enlarged.

If the processing line is curved, the extraction process of spatial frequency components is performed on the unit image that has been converted to a straight line through the coordinate conversion process described above. If the processing line is not curved, the extraction process of spatial frequency components is performed on the unit image that has not been converted to a straight line through the coordinate conversion process described above.

Spatial frequency components refer to stripe patterns that are repeated at regular intervals in a unit image, and the intervals correspond to the frequency. Stripe patterns that are repeated at regular intervals appear in the unit images as a result of machining marks (tool marks) remaining on the machined surface, so the spatial frequency components in the image are an index that represents the spacing and size (height of the unevenness) of the machining marks.

In the extraction process of spatial frequency components, the image processing unit 22 performs FFT (Fast Fourier Transform) processing on the unit images D3 and D4 as shown in FIG. 3 to obtain two-dimensional images F3 and F4 in which the spatial frequency components are shown in a two-dimensional graph. The two-dimensional images F3 and F4 are images showing two-dimensional graphs, with the center point O indicating a frequency zero component, and each point {distance r, angle θ} from the center point O indicating the frequency f and angle θ of the spatial frequency component (striped pattern). The amount of spatial frequency components is represented by the brightness of the two-dimensional images F3 and F4. As an example, the two-dimensional image F3 in FIG. 3 includes two points P1 and P2 with high brightness other than the zero point, indicating that the unit image D3 mainly includes two spatial frequency components. Also, the two-dimensional image F4 in FIG. 3 does not include any points with high brightness other than the zero point, indicating that the unit image D4 does not include a large spatial frequency component. Note that the two-dimensional images F3 and F4 in Figure 3 are shown with light and dark inverted.

In the extraction process of spatial frequency components, the image processing unit 22 may obtain the two-dimensional images F3 and F4, perform a filter process to extract parts with high brightness, and then perform IFFT (Inverse Fast Fourier transform) processing on the two-dimensional images F3 and F4 after the filter process. The IFFT process obtains unit images D3i and D4i of the workpiece K to be processed, in which only the spatial frequency components are prominently extracted. As an example, the unit image D3i in FIG. 3 is an image that includes stripe patterns J1 and J2 of two spatial frequencies corresponding to two points P1 and P2 in the two-dimensional image F3. Therefore, it is easy to determine that the original unit image D3 includes two types of processing marks corresponding to the two stripe patterns J1 and J2 from the unit image D3i. Moreover, the unit image D4i in FIG. 3 is an image that does not include stripe patterns, corresponding to the absence of large spatial frequency components other than the zero point. Therefore, it is easy to determine that almost no processing marks remain in the original unit image D4 from the unit image D4i.

<AnoGAN Processing Unit>
4 is a diagram for explaining the learning process (A) and the judgment process (B) executed by the AnoGAN processing unit. In general, AnoGAN (Anomaly Detection with Generative Adversarial Networks) refers to anomaly judgment in which a machine learning model having a structure called GAN is trained by providing a large number of normal images as training data, and an image to be judged is provided to the machine learning model after learning, thereby detecting anomalies in the image.

The AnoGAN processing unit 25 of this embodiment includes a machine learning model 25a having a GAN structure constructed in a computer, as shown in FIGS. 4A and 4B. As shown in FIG. 4A, the machine learning model 25a is given a large number of unit images Dtr cut out from a photographed image Qtr of a normally processed work surface as training data, and has already learned the training data. As shown in FIG. 4B, when each of a plurality of unit images D6 cut out from a photographed image Q6 of a workpiece K to be processed after processing is given, the machine learning model 25a extracts an abnormal portion an of each unit image D6 by AnoGAN. The unit images Dtr, which are training data, are selected from unit images of a processed workpiece that do not require reprocessing, and the machine learning model 25a determines a portion that may require reprocessing as an abnormal portion an. On the other hand, the determined abnormal portion an may be an abnormality based on other factors, not just a portion that requires reprocessing. Therefore, if the number (e.g. area) of abnormal areas an contained in unit image D6 is equal to or greater than a certain amount, it can be considered that there is a certain or greater possibility that unit image D6 contains a surface condition that requires reprocessing. On the other hand, if the number of abnormal areas an contained in unit image D6 is less than a certain amount, it can be considered that there is a less than a certain amount possibility that unit image D6 contains a surface condition that requires reprocessing.

The AnoGAN processing unit 25 further includes a sorting processing unit 25b that counts the total amount (e.g., total area) of abnormal areas an determined by the machine learning model 25a for the unit images D6 and compares the total amount of abnormal areas an with a predetermined threshold to sort the unit images D6. The AnoGAN processing unit 25 then sorts the unit images D6 whose comparison result is greater than the threshold as unit images D6a that have a certain or higher probability of being determined to require reprocessing, and the unit images D6 whose comparison result is less than the threshold as unit images D6na that have a less than certain probability of being determined to require reprocessing. The set of unit images D6na corresponds to an example of a first image group according to the present invention. The set of unit images D6a corresponds to an example of a second image group according to the present invention.

<Support Processing Unit and Surface Condition Designation Processing>
Fig. 5 is a flow chart showing a process for designating a surface state requiring re-processing, which is executed by the support processing unit, Fig. 6 is a display image diagram showing an example of a designation screen for designating a surface state requiring re-processing.

The support processing unit 21 starts the surface condition designation process of FIG. 5 when, for example, a user performs an operation to start the surface condition designation process. When the designation process starts, the support processing unit 21 first notifies the user to prepare a workpiece Ka including a portion that requires reprocessing (step S1). Here, the user prepares a workpiece Ka including a sufficiently processed portion and an insufficiently processed portion by applying the processing device 11 in various ways to the workpiece Ka for trial processing. Alternatively, a workpiece Ka that has been processed in various ways, such as a sufficiently processed portion and an insufficiently processed portion, may be prepared separately. Then, when the user notifies the support processing unit 21 that the workpiece Ka has been placed on the workbench (YES in step S2), the support processing unit 21 drives the photographing unit 12 to photograph the surface of the workpiece Ka (step S3).

Next, the support processing unit 21 divides the photographed image acquired in step S3 into unit images (step S4), passes the unit images to the image processing unit 22, and causes the image processing unit 22 to perform the coordinate conversion process, FFT process, and IFFT process described above on the unit images (steps S5 and S6). Through the processes of steps S5 and S6, a plurality of unit images are obtained in which the stripe pattern contained in the surface of the workpiece Ka has been extracted, as shown in the IFFT image in FIG. 3.

Then, the support processing unit 21 performs a display process to present the unit images obtained in steps S5 and S6 on the screen of the display unit 15A and to allow the user to specify the surface condition that requires reprocessing (step S7). More specifically, as shown in FIG. 6, the display unit 15A outputs a list display H1 of the multiple unit images obtained in steps S5 and S6, and a message H2 that prompts the user to specify the one that requires reprocessing from the list. In the example of FIG. 6, the list display H1 includes multiple sets of images, each set including the original unit image Dx, the two-dimensional image Fx after FFT processing, and the unit image Dxi after IFFT processing. Note that the list display H1 may include only the unit image Dxi after IFFT processing, or may include only the two-dimensional image Fx after FFT processing, or may include a combination of these and the unit image Dx before FFT processing.

Then, the support processing unit 21 determines whether the image designation is complete (step S8) and repeats the determination in step S8 until the image designation is complete.

Here, the user can easily identify how much fineness of the processing marks remain by paying attention to the information on the spatial frequency components of the processed surface shown in the two-dimensional image Fx after FFT processing or the unit image Dxi after IFFT processing in the list display H1. Then, the user can determine whether each two-dimensional image Fx or unit image Dxi is in a surface state that requires re-processing or does not require re-processing based on the fineness and size of the processing marks (height of the unevenness). Here, for example, assume a case in which the workpiece K to be processed has rough initial processing marks before processing, and the rough processing marks are removed by processing and other fine processing marks are added. In such a case, the user can determine that the two-dimensional image Fx or unit image Dxi in which the spatial frequency components corresponding to the initial rough processing marks remain is a surface state that requires re-processing, and that the two-dimensional image Fx or unit image Dxi in which the spatial frequency components corresponding to the rough processing marks remain is a surface state that does not require re-processing, regardless of whether or not there are spatial frequency components corresponding to fine processing marks. The user then selects all images from the list H1 that have been determined to have a surface condition requiring reprocessing by performing an instruction operation via the input device 15B. Once the selection is complete, such as by operating the OK button H3 indicating completion, the support processing unit 21 proceeds to the next step.

Next, when the process proceeds, the support processing unit 21 compares the specified two-dimensional image Fx or unit image Dxi with the unspecified two-dimensional image Fx or unit image Dxi, and calculates the conditions for determining whether or not reprocessing is necessary by comparing what spatial frequency components are included and determining that the surface condition requires reprocessing (step S9). Specifically, for example, if an image (two-dimensional image Fx or unit image Dxi) containing high spatial frequency components (fine stripes) at a size equal to or larger than a predetermined amount is included in the unspecified image, the support processing unit 21 determines that the high spatial frequency components are tool eyes that do not require reprocessing. Also, for example, if an image containing low spatial frequency components (coarse stripes) at a size equal to or larger than a predetermined amount is included in the specified image, while such an image is not included in the unspecified image, the support processing unit 21 determines that the low spatial frequency components are tool eyes that require reprocessing. Furthermore, if an image containing less than a predetermined amount of the above-mentioned low spatial frequency components is included in the unspecified images, the support processor 21 may calculate a threshold value for the amount of components that determines whether or not reprocessing is required, based on the amount of low spatial frequency components contained in the image. Through such processing, the support processor 21 can determine conditions such as, for example, if a specific spatial frequency component is contained in an amount equal to or greater than a threshold, it is a part that requires reprocessing.

When the conditions for determining whether reprocessing is necessary are requested, the support processing unit 21 passes the conditions to the reprocessing determination unit 24 (step S10) and ends the designation process in FIG. 5.

<Processing control unit, re-processing judgment unit, and automatic processing>
6 is a flowchart showing the automatic machining process executed by the machining control unit and the re-machining determination unit. The automatic machining process is started when the workpiece K to be machined is set on the workbench and a machining start command is input by the user.

When the automatic machining process is started, the machining control unit 23 generates a movement path for the head unit 11a that performs machining on the machining target surface of the workpiece K (step S21). Here, the movement path may be provided to the machining control unit 23 by the user through direct teaching. Direct teaching corresponds to a method in which the user operates the robot arm 11b to move the head unit 11a along a desired path, thereby causing the machining control unit 23 to memorize the path. In addition, the user may calculate the movement path by providing the machining control unit 23 with position information and operation commands that determine the movement path, or the machining control unit 23 may calculate the movement path along the machining target surface based on the position detected by detecting the position of the workpiece K to be machined.

Then, the machining control unit 23 drives the head unit 11a while driving the robot arm 11b so that the head unit 11a moves along the generated movement path, thereby performing machining (step S22). This driving causes machining to be performed on the machining target surface of the workpiece K to be machined.

Once the processing is completed, the re-processing determination unit 24 drives the photographing unit 12 to photograph the entire surface of the workpiece K to be processed (step S23). The re-processing determination unit 24 then links the photographed image to location data indicating which location of the workpiece K it is, and divides the photographed image into multiple unit images (step S24).

Next, the reprocessing determination unit 24 sends the multiple unit images to the AnoGAN processing unit 25, which classifies the unit images into those that are more than a certain probability that the surface condition that should be determined to require reprocessing is included and those that are less than a certain probability that the surface condition that should be determined to require reprocessing is included based on the amount of detected abnormal areas an (step S25).

Then, the re-machining determination unit 24 sends the unit images that are more than a certain probability of including a surface state that should be determined to require re-machining to the image processing unit 22, which performs a coordinate transformation process (step S26) to convert curved machining marks into straight lines, and a process to extract spatial frequency components contained in the unit images by FFT processing or FFT processing and IFFT processing, and to digitize the extracted spatial frequency components into frequency and the magnitude of the frequency components (step S27). The frequency and the magnitude of the frequency components can be obtained by searching for positions of high brightness in the two-dimensional image after FFT processing, and from the position and brightness. Alternatively, the frequency and the magnitude of the frequency components can be obtained by determining the spacing of the stripes and the shading ratio of the stripes from the unit images after IFFT processing by image recognition processing.

The re-processing determination unit 24 then applies the values (frequency and magnitude of the frequency components) indicating the spatial frequency components extracted in step S27 to the conditions for determining whether re-processing is required that were passed from the support processing unit 21 in the designation process of FIG. 5, and determines whether re-processing is required (step S28). Here, the re-processing determination unit 24 performs the above-mentioned determination process for all unit images that have been subjected to the processes of steps S26 and S27.

As a result, it is determined whether there is a unit image that is determined to match the conditions (requires remachining) (step S29), and if the result is NO, the machining control unit 23 ends the automatic machining process for one workpiece K to be machined. On the other hand, if the result of the determination in step S29 is YES, the machining control unit 23 generates a movement path for moving the head unit 11a so as to remachine the part of the workpiece K to be machined that is shown in the unit image based on the part data linked to the unit image that is determined to match (step S30). Then, the machining control unit 23 drives the robot arm 11b while driving the head unit 11a so that the head unit 11a moves according to the generated movement path (step S31). The processing in step S31 applies remachining to the part that requires remachining.

After the re-processing in which the head unit 11a moves along the path generated in step S30 is completed, the processing control unit 23 and the re-processing determination unit 24 then repeat the process from step S23 until step S29 is determined to be NO (no unit image is determined to meet the conditions requiring re-processing). Then, if step S29 is determined to be NO, the automatic processing for processing one workpiece K to be processed is terminated. Through this automatic processing, the processing system 1 can process the workpiece K to be processed until there are no more areas requiring re-processing.

As described above, according to the machining system 1 of this embodiment, the support processing unit 21 presents information on multiple types of workpiece surfaces to the user, and the user can specify one of these pieces of workpiece surface information using the input device 15B. Then, in the subsequent automatic machining process, the remachining determination unit 24 determines the locations of the workpiece K that require remachining based on information obtained from the photographed image of the surface of the workpiece K after machining and the workpiece surface information specified by the user. Therefore, even if there are surface conditions that require remachining and surface conditions that do not require remachining, and the distinction between these varies depending on, for example, the product specifications, according to the machining system 1 of this embodiment, the locations that require remachining can be appropriately determined and the locations can be controlled so that they are remachined.

On the surface of the workpiece K to be machined, a pattern in which a similar pattern (e.g., unevenness) is repeated at regular intervals, such as by machining marks, may occur, and such patterns may have different fineness and different sizes (height of unevenness). In addition, it may be required to determine how many patterns of a certain fineness remain in a certain area and whether re-machining of that area is required. Therefore, according to the machining system 1 of this embodiment, information indicating the spatial frequency components of the workpiece surface is applied as information on the workpiece surface presented to the user in the surface condition designation process. By showing the information on the spatial frequency components, it is possible to eliminate shadows in the photograph, patterns such as unevenness that are not repeated, or noise in the image, and to highlight only the information on the pattern in which the same pattern is repeated at regular intervals, allowing the user to recognize it. Therefore, based on the information on the spatial frequency components, the user can accurately distinguish between a surface condition that requires re-machining and a surface condition that does not require re-machining, and can accurately designate these.

Furthermore, according to the machining system 1 of this embodiment, during automatic machining, the remachining determination unit 24 acquires spatial frequency component information from the captured image of the workpiece K to be machined after machining, and compares this information with spatial frequency component information specified by the user to determine the areas that require remachining. Through this processing, it is possible to detect with high accuracy the areas that require remachining according to the presence or absence or amount of a pattern that is repeated at regular intervals, such as machining marks, and to appropriately perform remachining of the workpiece K to be machined.

Furthermore, according to the processing system 1 of this embodiment, the image processing unit 22 performs FFT processing or FFT processing and IFFT processing on the captured image (its unit image) to generate information on the spatial frequency components contained in the captured image. By searching for points with high brightness contained in the two-dimensional image using the two-dimensional image obtained by FFT processing, it is possible to identify the spatial frequency and the magnitude of its components, and to identify patterns that are repeated at regular intervals, such as processing marks, with high accuracy. In addition, the unit image obtained by IFFT processing allows the user to easily grasp patterns that are repeated at regular intervals, such as processing marks, from the spacing and shading of the stripes.

Furthermore, according to the machining system 1 of this embodiment, the image processing unit 22 performs coordinate conversion processing on the captured image (its unit image), thereby converting the machining marks included in the captured image so that they become linear. Therefore, even if the workpiece K to be machined has a curved surface and the machining marks are captured in a curved shape depending on the direction in which the part is photographed, the influence of such a curved surface can be eliminated. As a result, the user can easily and accurately specify surface conditions that require remachining and surface conditions that do not require remachining, and the remachining determination unit 24 can determine with high accuracy the parts that require remachining.

Furthermore, according to the machining system 1 of this embodiment, the AnoGAN processing unit 25 divides the captured image (multiple unit images D6) of the workpiece K to be machined into unit images D6a whose abnormality determination area is equal to or greater than the threshold, and unit images D6na whose abnormality determination area is less than the threshold. Furthermore, the re-machining determination unit 24 does not determine whether re-machining is necessary for one of the divided unit images D6na, but determines whether re-machining is necessary for the other divided unit image D6a. Therefore, the number of unit images for which the re-machining determination unit 24 performs the determination process can be reduced, thereby reducing the load on the re-machining determination unit 24 and speeding up the process of determining whether re-machining is necessary.

The above describes an embodiment of the present invention. However, the present invention is not limited to the above embodiment. For example, in the above embodiment, a processing system is shown in which the control device 20 controls the processing device 11 to automatically process and reprocess the workpiece K to be processed. However, the processing and reprocessing may be performed by an operator operating the robot arm 11b, or may be performed manually by an operator. In this case, the processing control unit 23 is removed from the control device 20 in FIG. 1. Also, a configuration in which the processing control unit 23 and the processing device 11 are removed from the processing system 1 constitutes an example of a processing support device E according to the present invention.

In the above embodiment, the support processing unit 21 presents the user with a two-dimensional image that has been subjected to FFT processing or a unit image that has been subjected to IFFT processing in the surface condition designation process. However, when the two-dimensional image that has been subjected to FFT processing contains multiple spatial frequency components, or when the unit image that has been subjected to IFFT processing contains multiple spatial frequency components, the image may be divided into individual spatial frequency components and presented to the user. For example, when the two-dimensional image that has been subjected to FFT processing has a luminance peak at the first point P1 and the second point P2, the image may be divided into two images, one in which the peak at the first point P1 is left and the peak at the second point P2 is excluded, and the other in which the peak at the first point P1 is excluded and the peak at the second point P2 is left, and presented to the user. Alternatively, IFFT processing may be performed on each of the two images thus divided, and the two unit images obtained by this processing (a unit image in which a striped pattern corresponding to the peak at the first point P1 appears, and a unit image in which a striped pattern corresponding to the peak at the second point P2 appears) may be presented to the user. In the above embodiment, the support processing unit 21 includes a configuration in which the user specifies the surface condition that requires reprocessing, but the support processing unit 21 may already hold information on the surface condition that requires reprocessing in advance, or may hold the information by specification or learning from a time before that point in time. In this case, the support processing unit 21 does not need to have a configuration in which an image is presented to prompt the user to specify the surface condition.

In the above embodiment, the processing system for polishing was described, but the processing system according to the present invention is not limited to polishing, and can be applied to systems for various processes such as cutting, grinding, and special processes for processing the surface of a workpiece using discharge, laser irradiation, or electrolytic action. In the above embodiment, a configuration was shown in which a process for extracting spatial frequency components of a captured image (extracting striped processing marks) is incorporated into a process for specifying a surface condition and a process for determining whether or not reprocessing is required, taking as an example a polishing process in which striped processing marks are used as a basis for determining whether or not reprocessing is required. However, processes suitable for incorporating a process for extracting spatial frequency components of a captured image are not limited to polishing. For example, in a cutting process in which burrs are generated at approximately equal intervals, or other processes in which a surface condition in which a similar pattern appears repeatedly at a predetermined interval is used as a basis for determining whether or not reprocessing is required, incorporating a process for extracting spatial frequency components of a captured image is as useful as the above embodiment.

In the above embodiment, the surface condition designation process is described as a process in which the user designates surface information that requires reprocessing, but the process may be a process in which the user designates surface information that does not require reprocessing. In addition, the coordinate change process for converting curved processing marks in the captured image into straight lines, the distribution process of the captured image by AnoGAN, or both of these may be omitted. In addition, in the above embodiment, a configuration is shown in which the user is presented with information on spatial frequency components extracted from the captured image in the surface condition designation process, but the information on the work surface presented to the user may be a captured image showing the processed state as it is, or may be an image from which shadows and noise on the image have been removed. In addition, the details shown in the embodiment may be changed as appropriate within the scope of the invention.

REFERENCE SIGNS LIST 1 Processing system E Processing support device 11 Processing device 11a Head unit 11b Robot arm 12 Photography unit 15 Setting device 15A Display (presentation unit)
15B Input device (designation unit)
20 Control device 21 Support processing unit 22 Image processing unit 23 Machining control unit 24 Remachining determination unit 25 AnoGAN processing unit 25a Machine learning model 25b Allocation processing unit K Work to be machined D3, D4 Unit image (photographed image)
F3, F4 2D images (images after FFT processing)
D3i, D4i unit images (images after IFFT processing)
H1 List display

Claims (8)

A presentation unit that presents information on a plurality of types of work surfaces;
A designation unit capable of externally designating one of the multiple types of work surface information presented by the presentation unit;
An imaging unit for acquiring an image of a workpiece to be processed after the workpiece is processed;
a re-machining determination unit that determines a portion of the workpiece that needs to be re-machined based on information designated by the designation unit and information obtained from the captured image;
Equipped with
The information designated by the designation unit is designated as information indicating a surface condition requiring reprocessing,
A machining support device , wherein the information on the workpiece surface presented by the presentation unit is information indicating spatial frequency components of the workpiece surface . A presentation unit that presents information on a plurality of types of work surfaces;
A designation unit capable of externally designating one of the multiple types of work surface information presented by the presentation unit;
An imaging unit for acquiring an image of a workpiece to be processed after the workpiece is processed;
a re-machining determination unit that determines a portion of the workpiece that needs to be re-machined based on information designated by the designation unit and information obtained from the captured image;
Equipped with
The information designated by the designation unit is designated as information indicating a surface condition requiring reprocessing,
A processing support device , wherein the information obtained from the captured image is information indicating the spatial frequency components of the workpiece surface in the captured image . A presentation unit that presents information on a plurality of types of work surfaces;
A designation unit capable of externally designating one of the multiple types of work surface information presented by the presentation unit;
An imaging unit for acquiring an image of a workpiece to be processed after the workpiece is processed;
a re-machining determination unit that determines a portion of the workpiece that needs to be re-machined based on information designated by the designation unit and information obtained from the captured image;
Equipped with
The information designated by the designation unit is designated as information indicating a surface condition requiring reprocessing,
The photographing unit acquires the photographed images of a plurality of locations of the workpiece to be machined,
The image processing device further includes an AnoGAN processing unit that divides the captured images of the plurality of locations into a first image group and a second image group having a larger number of areas detected as abnormal than the first image group by AnoGAN,
The re-processing determination unit does not determine a portion of the captured images of the first image group that requires re-processing, and determines a portion of the captured images of the second image group that requires re-processing.
Processing support equipment. An imaging unit for acquiring an image of a workpiece to be processed after the workpiece is processed;
a re-machining determination unit that determines a portion of the workpiece that requires re-machining based on information indicating a surface state that requires re-machining and information obtained from the captured image;
Equipped with
A processing support device , wherein the information obtained from the captured image is information indicating the spatial frequency components of the workpiece surface in the captured image . An imaging unit for acquiring an image of a workpiece to be processed after the workpiece is processed;
a re-machining determination unit that determines a portion of the workpiece that requires re-machining based on information indicating a surface state that requires re-machining and information obtained from the captured image;
Equipped with
The photographing unit acquires the photographed images of a plurality of locations of the workpiece to be machined,
The image processing device further includes an AnoGAN processing unit that divides the captured images of the plurality of locations into a first image group and a second image group having a larger number of areas detected as abnormal than the first image group by AnoGAN,
The re-processing determination unit does not determine the locations of the captured images in the first image group that require re-processing, but determines the locations of the captured images in the second image group that require re-processing . A first image processing unit performs FFT or FFT and IFFT processing on the captured image,
information indicating the spatial frequency components is generated by the processing of the first image processing unit;
The machining support device according to claim 2 or 4 . A second image processing unit performs coordinate conversion so that curved processing marks included in the captured image become straight lines,
The first image processing unit performs the processing on the image that has been coordinate-transformed by the second image processing unit.
The machining support device according to claim 6 . A processing device that performs processing;
The machining support device according to any one of claims 1 to 7,
A processing system comprising:

Images