Inspection image processing device and inspection image processing method

Abstract

This inspection image processing device comprises an acquisition device and an arithmetic device. The acquisition device acquires an inspection image generated by imaging or measuring a structure in which a plurality of tape-shaped material pieces are arranged side by side. The arithmetic device generates an edge image in which the edge of the material pieces is emphasized by applying, to the inspection image, an edge-emphasizing filter that emphasizes an edge of the material pieces included in the inspection image, the filter emphasizing the edge at a position where an end of the material piece is assumed to exist on the basis of the design information of the structure.

Description

Inspection image processing apparatus and inspection image processing method

[0001] The present application mainly relates to an inspection image processing apparatus for inspecting a structure in which a plurality of strip-shaped material pieces are arranged side by side.

[0002] Patent Document 1 discloses a laminating apparatus that laminates prepregs as material pieces to form a laminated structure. The laminating apparatus includes an imaging device and an image processing apparatus. The imaging device captures images of the gaps between the prepregs. The image processing apparatus analyzes the captured image and measures the gaps between the prepregs. The laminating apparatus stops lamination when the measured value of the gap between the prepregs exceeds an allowable value.

[0003] Japanese Patent Laid-Open No. 6-55656

[0004] Material pieces such as prepreg may be arranged without gaps. In this case, it may be difficult to recognize the gaps between the material pieces or the edges of the material pieces from an image obtained by imaging or measurement. Furthermore, even if there is a gap between the material pieces, it may be difficult to recognize the gap between the material pieces or the edges of the material pieces depending on the imaging environment, the surface properties of the material pieces, or other factors. As a result, there are cases where a structure in which material pieces are arranged side by side cannot be accurately inspected using images, and improvements are needed.

[0005] The present application has been made in view of the above circumstances, and a main object thereof is to provide an image processing apparatus for accurately inspecting a structure in which material pieces are arranged side by side.

[0006] The problems to be solved by the present application are as described above. Next, the means for solving these problems and the effects thereof will be described.

[0007] According to a first aspect of this application, an inspection image processing device having the following configuration is provided. That is, the inspection image processing device comprises an acquisition device and a calculation device. The acquisition device acquires an inspection image generated by imaging or measuring a structure in which a plurality of tape-shaped material pieces are arranged side by side. The calculation device applies an edge enhancement filter to the inspection image, which is a filter that enhances the edges of the material pieces included in the inspection image, and is a filter that enhances the edges at positions where the edges of the material pieces are assumed to exist based on the design information of the structure, to generate an edge image in which the edges of the material pieces are enhanced.

[0008] According to a second aspect of this application, the following inspection image processing method is provided. That is, an inspection image is generated by imaging or measuring a structure in which a plurality of tape-shaped material pieces are arranged side by side. An edge enhancement filter is applied to the inspection image, which is a filter that enhances the edges of the material pieces included in the inspection image, and is a filter that enhances the edges at positions where the edges of the material pieces are assumed to exist based on the design information of the structure, thereby generating an edge image in which the edges of the material pieces are enhanced.

[0009] According to this application, a structure in which material pieces are arranged in a line can be inspected with high precision.

[0010] A diagram illustrating the configuration of a stacked system including an inspection image processing apparatus according to one embodiment of this application. A flowchart showing the process of inspecting a stacked structure. A diagram showing examples of normal and abnormal arrangement of material pieces. A flowchart showing image processing of the inspection image. A diagram showing a filter applied to the inspection image and its effect. A diagram showing the calculation of integrated data by adding up pixel values ​​along an ideal angle.

[0011] Next, embodiments of this application will be described with reference to the drawings. First, the lamination system 1 will be described with reference to Figure 1.

[0012] The lamination system 1 stacks material pieces 33 to create a laminated structure and also checks whether the material pieces 33 are properly arranged in the laminated structure.

[0013] A workbench 31 is provided in the workshop. A mold 32 is placed on the workbench 31. The mold 32 is a component for arranging material pieces 33. The mold 32 has a shape corresponding to the laminated structure.

[0014] The material piece 33 is, for example, a reinforcing fiber such as carbon fiber, glass fiber, or aramid fiber. The material piece 33 is a flexible, tape-shaped member. "Tape-shaped" means that it is thin and flat. In this embodiment, the material piece 33 is a long, straight member, but the material piece 33 may also include a curved shape. The resulting laminated structure may be used, for example, as a panel material for vehicles such as aircraft or automobiles, but it may also be used for other purposes.

[0015] As shown in Figure 1, the stacking system 1 comprises a robot 11, a robot controller 12, a laser sensor 21, a camera 22, and a sensor controller 23.

[0016] The robot 11 comprises multiple arms and multiple actuators. The multiple arms are rotatably connected to each other. The actuators are, for example, motors, which generate power to rotate the arms. Tools for the robot 11 to perform tasks are attached to the ends of the arms.

[0017] The robot controller 12 comprises a processor and storage. The processor is, for example, a CPU, which performs arithmetic processing. The storage is, for example, an HDD, SSD, or flash memory, which stores programs and control data. The robot controller 12 controls the robot 11 by having the processor read and execute a program from the storage. Specifically, the robot controller 12 controls the position and orientation of the arm or tool by transmitting commands to the actuator.

[0018] The laser sensor 21 detects the surface shape of the stacked structure by irradiating it with a laser, receiving the reflected light, and analyzing it. Based on the detection result of the laser sensor 21, a distance image is generated. A distance image is an image in which distance information is contained for each pixel. In this embodiment, the distance information is the distance from the laser sensor 21, or the height from the reference surface of the stacked structure. The camera 22 generates an image including the stacked structure by imaging the stacked structure. The image generated by the camera 22 is called a color image, but may also be called an RGB image or an optical image. For inspection of the stacked structure, only the distance image and / or the color image may be used, or both may be used. Hereinafter, the general term for images used for inspection of the stacked structure is defined as "inspection image".

[0019] The sensor controller 23 generates control signals to be transmitted to the laser sensor 21 or camera 22, and processes detection results input from the laser sensor 21 or camera 22. The hardware configuration of the sensor controller 23 is the same as that of the robot controller 12, so a detailed explanation is omitted.

[0020] The stacking system 1 includes an inspection image processing device 40 for inspecting the stacked structure based on inspection images. The inspection image processing device 40 is, for example, an information processing device such as a PC on which an inspection program is installed. The inspection image processing device 40 includes an acquisition device 41 and a calculation device 42.

[0021] The acquisition device 41 is a communication module for communicating with the robot controller 12 and the sensor controller 23. The acquisition device 41 communicates with the robot controller 12 and the sensor controller 23 via a local area network or signal line. The inspection image processing device 40 may be located on the cloud. In this case, the acquisition device 41 further communicates with the robot controller 12 and the sensor controller 23 via a router and the internet, etc.

[0022] The arithmetic unit 42 comprises a processor and storage. The processor is, for example, a CPU, which performs arithmetic processing. The storage is, for example, an HDD, SSD, or flash memory, which stores programs and control data. In particular, the storage stores an inspection program for performing inspections using inspection images. The processor reads the inspection program from the storage and executes it, causing the inspection image processing device 40 to perform inspections using the inspection images.

[0023] Next, with reference to Figure 2, the overall process of laminating the material pieces 33 and inspecting the laminated structure will be described. Figure 2 is a flowchart showing the processes performed by the lamination system 1. In the flow of Figure 2, the processes related to the robot 11 are performed by the robot controller 12, the processes related to the laser sensor 21 or camera 22 are performed by the sensor controller 23, and the inspection processes are performed by the computing unit 42. Any of the above may be used to perform the remaining processes.

[0024] First, the lamination system 1 scans the mold 32 (S101). Specifically, the robot controller 12 controls the robot 11 to position the tip of the arm near the mold 32. In this state, the sensor controller 23 commands the laser sensor 21 to start measurement and the camera 22 to start imaging. While performing measurement with the laser sensor 21 and imaging with the camera 22, the robot controller 12 moves the tip of the arm along the surface of the mold 32. In this way, the mold 32 is scanned. Note that the scanning of the mold 32 may also be performed without using the robot 11. For example, the mold 32 can be scanned by fixing the laser sensor 21 or camera 22 above the workbench and performing measurement and imaging while moving the workbench table as needed.

[0025] Next, the lamination system 1 analyzes the three-dimensional shape of the mold 32 based on the information obtained by scanning (S102). The analysis results are transmitted to the inspection image processing device 40 or another higher-level control device. The inspection image processing device 40 or the other higher-level device creates shape data for each layer of material piece 33 based on the analysis results of the mold 32 and the information of the final lamination structure.

[0026] The following process is performed for each layer. The layer being processed is referred to as layer N (N = 1, 2, ...). First, the lamination system 1 acquires the shape data of the Nth layer material piece 33 (S103). Next, the lamination system 1 uses the robot 11 to position the material piece 33 based on the acquired shape data (S104). In this embodiment, the robot 11 is used to position the material piece 33, but an operator may also position the material piece 33 manually.

[0027] Next, the lamination system 1 uses the robot 11, laser sensor 21, and camera 22 to generate an inspection image of the Nth layer material piece 33 (S105). The method for generating the inspection image is the same as the method for scanning the mold 32. Next, the inspection image processing device 40 processes the inspection image and measures the position of the material piece 33 (S106). Details of the image processing by the inspection image processing device 40 will be described later.

[0028] Next, the lamination system 1 calculates the difference between the ideal design value of the material piece 33 and the measured value obtained by measurement (S107). The lamination system 1 determines whether this difference is less than or equal to the tolerance (S108). If the difference is greater than the tolerance, the lamination system 1 notifies an error (S109). In this case, the process is repeated. If the difference is less than or equal to the tolerance, the lamination system 1 determines whether the N layer is the final layer (S110). If the N layer is not the final layer, the lamination system 1 sets the next layer (S111) and repeats the process from steps S103 to S107. If the N layer is the final layer, the lamination system 1 terminates the process.

[0029] Next, with reference to Figure 3, the inspection of the arrangement of the material pieces 33 will be described in detail.

[0030] In the following explanation, the term "design information" will be used. Design information refers to information related to the design of a laminated structure, and in particular, information related to the material pieces 33. Design information includes not only the shape data (width, thickness, etc.) of a single material piece 33, but also the shape data of each layer created by arranging multiple material pieces 33 (the overall shape of one layer, or the coordinates of the material pieces 33 to be arranged, the arrangement angle of the material pieces 33, etc.). Furthermore, the term "based on design information" used below means that processing is performed using at least a part of the design information.

[0031] Figure 3 shows an example of the arrangement of material pieces 33. The material pieces 33 are arranged along a predetermined angle. More specifically, elongated or straight-shaped material pieces 33 are arranged along a predetermined angle. Hereinafter, the angle at which the material pieces 33 are arranged will be referred to as the arrangement angle, and the arrangement angle specified in the design information will be referred to as the ideal angle.

[0032] Figure 3 shows normal examples where the ideal angles are 0 degrees and 90 degrees. As shown in Figure 3, the material pieces 33 are normally arranged side by side without any gaps. Figure 3 shows abnormal examples where there is an angle misalignment and gaps. An angle misalignment means that the measured arrangement angle is significantly different from the ideal angle. Gaps mean that there is a large gap between the material pieces 33. Note that the abnormal examples shown in Figure 3 are just one example.

[0033] When inspecting the layered structure of material pieces 33 using images, it is necessary to identify the position of the material pieces 33. However, identifying the position of the material pieces 33 can be difficult. This is because the edges of material pieces 33, which are arranged without gaps, hardly appear in the inspection image. In contrast, in this embodiment, by performing a process to identify the edges of the material pieces 33, it is possible to perform a highly accurate inspection of the arrangement of the material pieces 33.

[0034] The process of measuring the position of the material piece 33 using inspection images will be described in detail below with reference to Figures 4 to 6. The flowchart in Figure 4 shows the process executed by the arithmetic unit 42 of the inspection image processing device 40.

[0035] First, the calculation unit 42 acquires an inspection image from the sensor controller 23 using the acquisition device 41 (S201). Next, the calculation unit 42 sets the shift angle θ (S202). The shift angle θ is an angle for evaluating the arrangement angle of the material piece 33. The angle θ includes 0 degrees and positive and negative angles for shifting the ideal angle. The specific shift angle θ is arbitrary as it depends on the accuracy required for the laminated structure, but for example, a value of 5 degrees or less, or 3 degrees or less can be used. The number of shift angles θ used in the processing is, for example, 3 to 10.

[0036] Next, the arithmetic unit 42 applies a change enhancement filter to the inspection image (S203). The change enhancement filter compares the target pixel with surrounding pixels, calculates the difference in pixel values, and enhances the target pixel based on the magnitude of the difference. The larger the difference, the higher the degree of enhancement. The pixel value indicates distance in a depth image and brightness in a color image. For example, if a pixel corresponding to the edge of the material piece 33 has a higher pixel value than surrounding pixels, the edge of the material piece 33 can be made to stand out. Note that the change enhancement filter does not selectively enhance the edge of the material piece 33, so pixels with higher pixel values ​​than surrounding pixels due to factors other than the edge are also enhanced.

[0037] A change enhancement filter is, for example, a Sobel filter, but any filter that highlights differences as described above may be used instead of the Sobel filter. For example, a Prewitt filter or a Laplacian filter can be used as a change enhancement filter. Figure 5 shows an example of image changes when a change enhancement filter is applied. Note that Figure 5 is a conceptual diagram to make the effects of each filter easier to understand and does not represent measured values. Note that applying a change enhancement filter is not mandatory and can be omitted.

[0038] Next, the arithmetic unit 42 applies an edge enhancement filter to the inspection image (S204). Unlike the change enhancement filter, the edge enhancement filter is a filter for selectively enhancing the edges of the material piece 33. The edge enhancement filter allows you to set conditions for enhancing pixel values ​​by adjusting its parameters. More specifically, the edge enhancement filter allows you to adjust the spacing at which pixels to be enhanced may occur and the direction in which the pixels to be enhanced are aligned. The width of the material piece 33 is set as the spacing, and the ideal angle + shift angle θ is set as the direction. This allows for selective enhancement of the edges of the material piece 33. Figure 5 shows an example of how the image changes when the edge enhancement filter is applied. Hereafter, the image to which the edge enhancement filter has been applied may be referred to as the edge image.

[0039] As a specific filter, for example, a Gabor filter can be used. A Gabor filter is a filter in which the frequency parameter and direction parameter can be adjusted. By adjusting the frequency parameter, pixels can be periodically enhanced. Therefore, by adjusting the frequency parameter to a value corresponding to the width of the material piece 33 and the direction parameter to a value corresponding to the ideal angle + shift angle θ, the edges of the material piece 33 can be selectively enhanced. Edge enhancement filters are not limited to Gabor filters. An edge enhancement filter may also be, for example, a wavelet filter.

[0040] The edge enhancement filter is a filter that is adjusted according to the width of the material piece 33 and the arrangement angle of the material piece 33. Here, the width and arrangement angle of the material piece 33 are the design information described above or information obtained by shifting said design information. In other words, the edge enhancement filter corresponds to a filter adjusted based on the design information. In this embodiment, since the shape and layout of the material piece 33 are predetermined, by adjusting the filter using this information, even objects that are normally difficult to detect can be enhanced and detected.

[0041] Next, the arithmetic unit 42 creates integrated data by accumulating pixel values ​​along the direction indicated by the ideal angle + shift angle θ (S205). Figure 6 shows an example where the ideal angle + shift angle θ is 0 degrees, so each pixel value is accumulated along the imaginary line of 0 degrees. The graph below shows the integrated data. The integrated data is data that shows the sum of pixel values ​​according to the position. The sum of pixel values ​​is the value obtained by accumulating pixel values ​​along the direction indicated by the ideal angle + shift angle θ as described above. As shown in the first example, when the degree of agreement between the ideal angle + shift angle θ and the edge direction is high, the parts with high and low pixel value sums are clearly separated. On the other hand, as shown in the second example, when the degree of agreement between the ideal angle + shift angle θ and the edge direction is low, differences in the sum of pixel values ​​become less likely to occur. For example, the standard deviation of the integrated data shows the degree of agreement between the ideal angle + shift angle θ and the edge direction. Note that the standard deviation is just one example, and it could also be the variance or the size of the interquartile range. In this way, the integrated data is evaluated.

[0042] Next, the calculation unit 42 determines whether all shift angles θ have been set (S206). If there are any shift angles θ that have not been set, the calculation unit 42 repeats the process from steps S202 to S205.

[0043] In this embodiment, the parameters of the edge enhancement filter are determined according to the set shift angle θ, and the summation direction when creating the integrated data is determined. In other words, different edge enhancement filters are applied and the summation direction differs depending on the set shift angle θ. Alternatively, the inspection image may be rotated according to the shift angle θ. In this case, a common edge enhancement filter can be used for all shift angles θ. The summation direction will also be the same for all shift angles θ. In other words, the difference is whether the shift angle θ is applied to the inspection image side or to the calculation side, but the processing performed is essentially the same.

[0044] When all shift angles θ have been set, the arithmetic unit 42 compares the integrated data, and acquires the shift angle θ and edge position with the maximum evaluation of the integrated data (S207). The comparison of integrated data is specifically a comparison of evaluation values indicating the standard deviation or the like obtained from the integrated data. A high evaluation of the integrated data means that the ideal angle + the shift angle θ is closest to the actual arrangement angle of the material piece 33. That is, the shift angle θ selected here indicates an approximate angle deviation amount. The edge position is a coordinate group of a plurality of pixels indicating one edge in the inspection image.

[0045] Next, the arithmetic unit 42 selects one edge position (S208). The arithmetic unit 42 generates a straight line along the selected edge position while removing outliers (S209). The edge position acquired in step S207 may include a position significantly different from the edge of the material piece 33. This is because pixel values at locations where no edge of the material piece 33 exists may increase due to noise, impurities, wrinkles of the material piece 33, unevenness of the laminated surface, and the like. Therefore, in step S209, instead of performing straight-line approximation on all pixel values, outliers are removed from all pixel values, and straight-line approximation is performed on the remaining pixel values. As is clear from the design information, since the edge of the material piece 33 has a shape along a straight line, the probability that a pixel located at a position significantly away from the straight line is an edge is very low. Therefore, parameters related to outlier removal are set from this perspective. This enables straight-line approximation after removing outliers in consideration of design information.

[0046] As a specific straight-line approximation method, for example, the RANSAC algorithm can be used. The RANSAC algorithm stochastically selects part of data, performs threshold determination, removes outliers, and performs straight-line approximation. It is necessary to perform threshold determination such that pixels far away from the edge of the material piece 33 become outliers, and it is preferable to set an error threshold corresponding thereto as a parameter.

[0047] Note that depending on the accuracy of the inspection image or the required measurement accuracy, the process of removing outliers is not an essential process and can be omitted. Further, instead of the process of removing outliers, a process of reducing the influence of outliers may be performed. The process of reducing the influence of outliers is, for example, a process of reducing the weight assigned to outliers.

[0048] Next, the arithmetic device 42 converts the linear expression obtained by linear approximation into an angle (S210). The angle obtained here is an arrangement angle with higher accuracy compared to the ideal angle + shift angle θ. In particular, the arrangement angle can be measured by accurately identifying even edges that cannot be identified only by applying a change enhancement filter to an inspection image. Therefore, high-precision inspection can be performed.

[0049] Next, the arithmetic device 42 determines whether all edge positions have been selected (S211). If the selection of all edge positions has not been completed, the arithmetic device 42 performs the processes from step S208 to S210 on another edge position. If the selection of all edge positions has been completed, the arithmetic device 42 outputs a result indicating the angle of each edge (S212).

[0050] The edge angles and linear expressions calculated here correspond to the measurement values in step S107, and are used for calculating differences. For example, the arithmetic device 42 calculates the difference between the edge angle and the ideal angle, and notifies an error if the difference exceeds an allowable error. Further, the linear expression of an edge is information indicating the measurement result of the position of the material piece 33. By comparing this linear expression of the edge with the edge position based on design information, the measured edge position is compared with the designed edge position, and the difference therebetween is calculated. The arithmetic device 42 notifies an error if the difference at the edge position exceeds the allowable error. As described above, in the present embodiment, the arrangement of the material piece 33 can be inspected based on edges enhanced using a change enhancement filter, an edge enhancement filter, or the like.

[0051] Preferred embodiments of the present application have been described above, but the above configuration can be modified, for example, as follows. Modifications may be made individually, or a plurality of modifications may be made in any combination.

[0052] In this embodiment, the inspection image processing device 40 processes the inspection image to measure the position of the material piece 33, etc., and then performs the inspection. Alternatively, the inspection image processing device 40 may only perform the process of measuring the position of the material piece 33, etc., by processing the image. In this case, the inspection image processing device 40 transmits the measurement results to another device, and that other device performs the inspection.

[0053] In this embodiment, a laser sensor 21 and a camera 22 are provided on the robot 11, but one of the laser sensor 21 or camera 22 may be omitted.

[0054] In this embodiment, the laminated structure of the material pieces 33 is the object of inspection. However, the laminated structure is just one example, and the image processing and inspection of this embodiment can also be applied to a structure in which only one layer of material pieces 33 or other components is arranged. Furthermore, even if the edges of the material pieces 33 have a gently curved shape, the processing of this application can be applied by approximating the curved shape as a set of multiple straight lines.

[0055] (Feature 1) The inspection image processing device 40 of this embodiment comprises an acquisition device 41 and a calculation device 42. The acquisition device 41 acquires an inspection image generated by imaging or measuring a structure in which a plurality of tape-shaped material pieces 33 are arranged side by side. The calculation device 42 applies an edge enhancement filter to the inspection image, which is a filter that enhances the edges of the material pieces 33 included in the inspection image, and is a filter that enhances the edges at positions where the ends of the material pieces 33 are assumed to exist based on the design information of the structure, to generate an edge image in which the edges of the material pieces 33 are enhanced.

[0056] Because it can generate edge images with precisely highlighted edges of material pieces, it can produce images suitable for high-precision inspection.

[0057] (Feature 2) In the inspection image processing device 40 of this embodiment, the arithmetic unit 42 calculates the difference in the value of the target pixel compared with the surrounding pixels, applies a change enhancement filter to the inspection image to enhance the target pixel based on the magnitude of the difference, and applies an edge enhancement filter to the inspection image to which the change enhancement filter has been applied.

[0058] Edges detected by the change enhancement filter can be enhanced by the edge enhancement filter. Therefore, even edges that cannot be identified at the stage of acquisition by the acquisition device 41 can be included in the edge image.

[0059] (Feature 3) In the inspection image processing device 40 of this embodiment, the material piece 33 has a shape that follows a straight line. The calculation device 42 applies an edge enhancement filter, whose parameters are set based on the arrangement angle of the material piece 33 and the width of the material piece 33 from the design information, to the inspection image to generate an edge image.

[0060] Since the position of the edge of the material piece 33 can be identified based on the angle at which the material piece 33 is positioned and the width of the material piece 33, the edge can be highlighted based on the design information.

[0061] (Feature 4) In the inspection image processing device 40 of this embodiment, the arithmetic unit 42 generates multiple patterns of edge images by rotating the inspection image, or by changing the value of the edge enhancement filter that corresponds to the arrangement angle.

[0062] By generating multiple edge image patterns as described above, it is possible to generate edge-emphasized images even when the placement angle of the material pieces deviates from the design information.

[0063] (Feature 5) In the inspection image processing device 40 of this embodiment, the calculation device 42 creates integrated data for estimating the edge position by summing the pixel values ​​of each pixel of the edge image in a direction along the arrangement angle of the material piece 33 included in the design information or an angle obtained by shifting the arrangement angle.

[0064] This allows us to create data useful for estimating edge locations.

[0065] (Feature 6) In the inspection image processing device 40 of this embodiment, the arithmetic unit 42 determines a shift angle θ for evaluation by shifting the arrangement angle. The arithmetic unit 42 generates an edge image in which an edge enhancement filter adjusted according to the shift angle θ is applied to the inspection image, or an edge image in which the edge enhancement filter is applied after rotating the inspection image according to the shift angle θ. The arithmetic unit 42 creates integrated data by summing the values ​​of each pixel in the edge image along the arrangement angle or the angle in which the arrangement angle is adjusted by the shift angle θ.

[0066] This allows us to measure whether or not the material pieces are positioned at a particular angle.

[0067] (Feature 7) In the inspection image processing device 40 of this embodiment, the calculation device 42 determines a plurality of shift angles θ. The calculation device 42 compares a plurality of the accumulated data adjusted according to the plurality of shift angles.

[0068] It is possible to measure which of several shift angles the angular displacement of a material piece is closest to.

[0069] (Feature 8) In the inspection image processing device 40 of this embodiment, the calculation device 42 estimates the position of the edge of the material piece 33 while performing a process to remove outliers from the data included in the edge image or a process to reduce the influence of outliers, based on the design information.

[0070] By reducing the influence of numerous noises contained in the edge image, the estimation accuracy of the edge position tends to be higher.

[0071] (Feature 9) In the inspection image processing device 40 of this embodiment, the calculation device 42 determines whether the arrangement of the material piece 33 is appropriate based on the estimated edge position of the material piece 33, the edge position of the material piece 33 included in the design information, and the tolerance.

[0072] The device can determine whether the position of the material pieces is correct without human intervention.

[0073] (Feature 10) In the inspection image processing device 40 of this embodiment, the calculation device 42 determines whether the arrangement angle of the material piece 33 is appropriate based on the arrangement angle of the material piece 33 obtained from the edge of the material piece 33, the arrangement angle included in the design information, and the tolerance.

[0074] The device can determine whether the placement angle of the material pieces is appropriate without human intervention.

[0075] An inspection image processing device can be realized by combining the above-mentioned features 1 to 10, for example, as follows. The same applies to the method or program. [Configuration 1] An inspection image processing device having feature 1 [Configuration 2] An inspection image processing device having feature 2 in addition to configuration 1 [Configuration 3] An inspection image processing device having feature 3 in addition to configuration 1 or 2 [Configuration 4] An inspection image processing device having feature 4 in addition to any one of configurations 1 to 3 [Configuration 5] An inspection image processing device having feature 5 in addition to any one of configurations 1 to 4 [Configuration 6] An inspection image processing device having feature 6 in addition to any one of configurations 1 to 5 [Configuration 7] An inspection image processing device having feature 7 in addition to any one of configurations 1 to 6 [Configuration 8] An inspection image processing device having feature 8 in addition to any one of configurations 1 to 7 [Configuration 9] An inspection image processing device having feature 9 in addition to any one of configurations 1 to 8 [Configuration 10] An inspection image processing device having feature 10 in addition to any one of configurations 1 to 9

[0076] The functions of the elements disclosed herein can be performed using circuits or processing circuits, including general-purpose processors, dedicated processors, integrated circuits, ASICs (Application Specific Integrated Circuits), conventional circuits, and / or combinations thereof, configured or programmed to perform the disclosed functions. A processor is considered a processing circuit or circuit because it includes transistors and other circuits. In this disclosure, a circuit, unit, or means is hardware that performs the enumerated functions, or hardware programmed to perform the enumerated functions. The hardware may be hardware disclosed herein, or other known hardware that is programmed or configured to perform the enumerated functions. If the hardware is a processor, which is considered a type of circuit, then the circuit, means, or unit is a combination of hardware and software, and the software is used to configure the hardware and / or the processor.

Claims

1. An inspection image processing apparatus comprising: an acquisition device that acquires an inspection image generated by imaging or measuring a structure in which a plurality of tape-shaped material pieces are arranged in a row; and a processing device that applies an edge enhancement filter, which is a filter that enhances the edges of the material pieces included in the inspection image and enhances the edges at positions where the edges of the material pieces are assumed to exist based on the design information of the structure, to the inspection image to generate an edge image in which the edges of the material pieces are enhanced.

2. An inspection image processing apparatus according to claim 1, wherein the arithmetic unit calculates the difference in the value of a target pixel compared with surrounding pixels, applies a change enhancement filter to the inspection image to enhance the target pixel based on the magnitude of the difference, and applies the edge enhancement filter to the inspection image to which the change enhancement filter has been applied.

3. An inspection image processing apparatus according to claim 1, wherein the material piece has a shape that follows a straight line, and the calculation device applies the edge enhancement filter, whose parameters are set based on the arrangement angle of the material piece and the width of the material piece from the design information, to the inspection image to generate the edge image.

4. An inspection image processing apparatus according to claim 3, wherein the arithmetic unit generates multiple patterns of the edge image by rotating the inspection image, or generates multiple patterns of the edge image by changing a value in the edge enhancement filter that corresponds to the arrangement angle.

5. An inspection image processing apparatus according to claim 1, wherein the arithmetic unit creates integrated data for estimating the edge position by adding up the values ​​of each pixel in the edge image in a direction along the arrangement angle of the material piece included in the design information or the angle obtained by adjusting the arrangement angle with a shift angle.

6. An inspection image processing apparatus according to claim 3, wherein the arithmetic unit determines a shift angle for evaluation by shifting the arrangement angle, the arithmetic unit generates an edge image obtained by applying the edge enhancement filter adjusted according to the shift angle to the inspection image, or an edge image obtained by rotating the inspection image according to the shift angle and then applying the edge enhancement filter, and the arithmetic unit creates integrated data for estimating the edge position by adding up the values ​​of each pixel in the edge image along the arrangement angle or the arrangement angle adjusted by the shift angle.

7. An inspection image processing apparatus according to claim 6, wherein the calculation device determines a plurality of shift angles, and the calculation device compares a plurality of the accumulated data adjusted according to the plurality of shift angles.

8. An inspection image processing apparatus according to claim 1, wherein the calculation device estimates the position of the edge of the material piece while performing a process to remove outliers from the data included in the edge image or a process to reduce the influence of outliers, based on the design information.

9. An inspection image processing apparatus according to claim 8, wherein the calculation device determines whether the arrangement of the material piece is appropriate based on the estimated edge position of the material piece, the edge position of the material piece included in the design information, and the tolerance.

10. An inspection image processing apparatus according to claim 8, wherein the calculation device determines whether the arrangement angle of the material piece is appropriate based on the arrangement angle of the material piece obtained from the edge of the material piece, the arrangement angle included in the design information, and the tolerance.

11. An inspection image processing method comprising: generating an inspection image by imaging or measuring a structure in which multiple tape-shaped material pieces are arranged in a row; and applying an edge enhancement filter to the inspection image, which is a filter that enhances the edges of the material pieces included in the inspection image, and is a filter that enhances the edges at positions where the edges of the material pieces are assumed to exist based on the design information of the structure, to generate an edge image in which the edges of the material pieces are enhanced.

Images