Workpiece image analysis device, method, and computer-readable medium

Abstract

The present invention aims to reduce the workload required to measure a workpiece. A workpiece image analysis device is provided with: a display section that displays an image of a workpiece that is a processing target of an industrial machine; an analysis object specification section that receives a specification of an analysis object for the image of the workpiece; a feature extraction section that extracts a structural feature of the analysis object specified by the analysis object specification section in a coordinate system of a real world; an analysis item setting section that receives a specification of an analysis item for the image of the workpiece; and a physical quantity calculation section that extracts a feature required for calculation of a physical quantity involved in the analysis item specified by the analysis item setting section from the structural feature of the analysis object, and calculates the physical quantity using the extracted feature.

Description

Technical Field

[0001] The present invention relates to a workpiece image analyzing device, a workpiece image analyzing method, and a computer-readable medium storing a program. Background Art

[0002] Conventionally, there is known a technique for measuring a workpiece to be processed for the purpose of performing processing using a machine tool.

[0003] When measuring workpieces, measurement methods using touch sensors (touch probes) or laser sensors generally offer advantages in terms of high resolution and accuracy. However, these methods have the disadvantages of narrowing the measurement range and increasing measurement time. Furthermore, when using touch sensors for measurement, for example, the operator manually moves the touch sensor, placing a significant workload on the operator to avoid damaging the workpiece or the touch sensor.

[0004] To reduce this workload, a system is known that automatically generates a measurement program that moves the touch sensor when the coordinates of the measurement point or adjacent points are input. However, grasping and inputting the coordinates of such multiple points while taking the coordinate system into consideration still places a heavy workload on the operator.

[0005] On the other hand, methods that use images obtained by visual sensors, etc. to measure the shape, position, etc. of a workpiece generally have the advantage of being able to measure a large area in a short time, but have the disadvantage of being impractical in preparation for machining, such as setting the workpiece coordinate system, from the perspective of measurement resolution and repeatability.

[0006] To address this issue, a workpiece measurement method has been proposed that combines workpiece imaging with measurement methods using touch sensors or laser sensors to complement each other's shortcomings.

[0007] For example, a known method displays a workpiece image captured by a visual sensor on a monitor, allows the user to specify a measurement point or adjacent points on the image through touch operations, and generates an automatic measurement program using a touch probe based on the coordinates of the points. For example, see Patent Document 1.

[0008] Prior art literature

[0009] Patent Literature

[0010] Patent Document 1: Japanese Patent Application Publication No. 2018-018155 Summary of the Invention

[0011] Problems to be solved by the invention

[0012] However, in the technology described in Patent Literature 1, depending on knowledge, experience, and the measurement object related to the setting of the measurement point, the adjacent point, or the like, sometimes complicated procedures of setting multiple measurement points, adjacent points, directions, and the like are required. Therefore, it is considered that higher convenience can be achieved if the measurement object can be intuitively specified with fewer operations.

[0013] Therefore, a technology that reduces the work burden required for measurement of a workpiece is desired.

[0014] Means for solving the problem

[0015] (1) One embodiment of a workpiece image analysis apparatus of the present disclosure includes: a display section that displays an image of a workpiece that is a processing object of an industrial machine; an analysis object specifying section that receives a specification of an analysis object with respect to the image of the workpiece; a feature extraction section that extracts a structural feature of the analysis object specified by the analysis object specifying section in a coordinate system of a real world; an analysis item setting section that receives a specification of an analysis item with respect to the image of the workpiece; and a physical quantity calculation section that extracts a feature required for calculation of a physical quantity related to an analysis item specified by the analysis item setting section from the structural feature of the analysis object, and calculates the physical quantity using the extracted feature.

[0016] (2) One embodiment of a workpiece image analysis method of the present disclosure includes: a display step of displaying an image of a workpiece that is a processing object of an industrial machine; an analysis object specifying step of receiving a specification of an analysis object with respect to the image of the workpiece; a feature extraction step of extracting a structural feature of the analysis object specified in the analysis object specifying step in a coordinate system of a real world; an analysis item setting step of receiving a specification of an analysis item with respect to the image of the workpiece; and a physical quantity calculation step of extracting a feature required for calculation of a physical quantity related to an analysis item specified in the analysis item setting step from the structural feature of the analysis object, and calculating the physical quantity using the extracted feature.

[0017] (3) One embodiment of a program of the present disclosure causes a computer to realize the following functions: a display control function that displays an image of a workpiece that is a processing object of an industrial machine; an analysis object specifying function that receives a specification of an analysis object with respect to the image of the workpiece; a feature extraction function that extracts a structural feature of the analysis object specified in the analysis object specifying function in a coordinate system of a real world; an analysis item setting function that receives a specification of an analysis item with respect to the image of the workpiece; and a physical quantity calculation function that extracts a feature required for calculation of a physical quantity related to an analysis item specified in the analysis item setting function from the structural feature of the analysis object, and calculates the physical quantity using the extracted feature.

[0018] Effects of the Invention

[0019] According to one embodiment, the workload required for measuring a workpiece can be reduced. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] Figure 1 This is a block diagram showing the configuration of a workpiece image analysis device according to one embodiment.

[0021] Figure 2 This is a schematic diagram showing an example of operations performed on the input screen for designating an analysis target.

[0022] Figure 3 This is a schematic diagram showing an example of operations performed on the input screen for designating an analysis target.

[0023] Figure 4 This is a schematic diagram showing an example of operations performed on the input screen for designating an analysis target.

[0024] Figure 5 This is a schematic diagram showing an example of operations performed on the input screen for designating an analysis target.

[0025] Figure 6 This is a schematic diagram showing an example of operations performed on the input screen for designating an analysis target.

[0026] Figure 7 This is a schematic diagram showing an example of a process of extracting a structural feature (three-dimensional shape) based on a designation operation input by a user.

[0027] Figure 8 This is a schematic diagram showing an example of a process of extracting a structural feature (three-dimensional shape) based on a designation operation input by a user.

[0028] Figure 9 This is a schematic diagram showing an example of a process of extracting a structural feature (three-dimensional shape) based on a designation operation input by a user.

[0029] Figure 10 This is a schematic diagram showing an example of a process of extracting a structural feature (three-dimensional shape) based on a designation operation input by a user.

[0030] Figure 11 This is a schematic diagram showing an example of a process of extracting a structural feature (three-dimensional shape) based on a designation operation input by a user.

[0031] Figure 12 This is a schematic diagram showing an example of conversion from the image coordinate system to the machine coordinate system.

[0032] Figure 13This is a schematic diagram showing an example of physical quantity calculation processing for a cylindrical workpiece.

[0033] Figure 14 This is a schematic diagram showing an example of physical quantity calculation processing for a cylindrical workpiece.

[0034] Figure 15 This is a schematic diagram showing an example of physical quantity calculation processing for a cylindrical workpiece.

[0035] Figure 16 This is a schematic diagram showing an example of physical quantity calculation processing for a rectangular parallelepiped workpiece having two holes formed therein.

[0036] Figure 17 This is a schematic diagram showing an example of physical quantity calculation processing for a rectangular parallelepiped workpiece having two holes formed therein.

[0037] Figure 18 This is a schematic diagram showing an example of physical quantity calculation processing for a rectangular parallelepiped workpiece having two holes formed therein.

[0038] Figure 19 This is a schematic diagram showing an example of physical quantity calculation processing for a cylindrical workpiece having a spherical shape.

[0039] Figure 20 This is a schematic diagram showing an example of physical quantity calculation processing for a cylindrical workpiece having a spherical shape.

[0040] Figure 21 This is a schematic diagram showing an example of physical quantity calculation processing for a cylindrical workpiece having a spherical shape.

[0041] Figure 22 This is a flowchart explaining the image analysis processing of the workpiece image analysis device. DETAILED DESCRIPTION

[0042] Hereinafter, one embodiment will be described with reference to the drawings.

[0043] <One embodiment>

[0044] Figure 1 This is a block diagram showing the configuration of a workpiece image analysis device 1 according to one embodiment.

[0045] The workpiece image analysis device 1 is composed of an information processing device such as a numerical controller or a PC (Personal Computer).

[0046] like Figure 1As shown, the workpiece image analysis device 1 includes a CPU (Central Processing Unit) 11 , a ROM 12 , a RAM 13 , an input unit 14 , a display unit 15 , a storage unit 16 , a communication unit 17 , and a visual sensor 18 .

[0047] Various system programs for controlling the workpiece image analysis device 1 , which are executed by the CPU 11 described later, are written in advance in the ROM 12 .

[0048] The RAM 13 is composed of a semiconductor memory such as a DRAM (Dynamic Random Access Memory), and stores data generated when the CPU 11 , which will be described later, executes various processes.

[0049] The input unit 14 is composed of an input device such as a keyboard, a mouse, or a touch sensor (touch panel), and receives input of various information from a user to the workpiece image analysis apparatus 1 .

[0050] The display unit 15 is composed of a display device such as an LCD (Liquid Crystal Display), and displays various processing results of the workpiece image analysis device 1 .

[0051] The storage unit 16 is composed of a non-volatile storage device such as a hard disk or flash memory, and stores programs for image analysis and the like. The storage unit 16 also stores an image database (image DB) 16a, which stores image data captured by a vision sensor 18 (described later), an analysis target feature database (analysis target feature DB) 16b, which stores features of analysis targets, and an analysis history database (analysis history DB) 16c, which stores past analysis history. The storage unit 16 may also store various processing results of the workpiece image analysis device 1.

[0052] The communication unit 17 includes a communication interface that performs signal processing based on a predetermined communication standard such as wired or wireless LAN, USB, etc., and controls communication between the workpiece image analysis apparatus 1 and other devices.

[0053] The visual sensor 18 includes a three-dimensional image capturing device such as a depth camera or a stereo camera, and captures a three-dimensional image of the workpiece to be analyzed. The captured three-dimensional image is an RGBD image (point cloud data) containing an RGB image and a range image D.

[0054] The CPU 11 controls the entire workpiece image analysis device 1 by executing various programs stored in the storage unit 16. For example, as described later, the CPU 11 executes a program for calculating physical quantities such as length, height, and diameter of various workpiece structures instructed by the user.

[0055] By executing the program for the calculation process, the CPU 11 forms a UI display control unit 11 a , an image acquisition unit 11 b , an analysis target designation unit 11 c , a feature extraction unit 11 d , an analysis item setting unit 11 e , and a physical quantity calculation unit 11 f as functional components.

[0056] <UI显示控制部11a>

[0057] During the calculation process of the physical quantity, the UI display control unit 11 a displays a user interface screen (UI screen) for the user to input and output various information.

[0058] For example, as described later, the UI display control unit 11a displays an input screen for accepting an instruction to obtain an image of a workpiece to be processed by a machine tool (not shown), or displays an input screen for specifying an analysis object in the obtained image of the workpiece, or displays an input screen for accepting a selection from candidates for an analysis item, or displays an analysis result of the specified analysis object.

[0059] Furthermore, when input is performed in the UI display control unit 11a, input can be received through a mouse, keyboard, touch operation, etc. For example, in addition to various input methods based on touch operation, it is also possible to draw an enclosing line using the arrow keys on the keyboard, draw a rectangular range by dragging the mouse, draw a point by pressing the Enter key on the keyboard or clicking the mouse, etc.

[0060] Furthermore, the workpiece is not limited to a workpiece to be processed by a machine tool (not shown), and can also be applied to a workpiece to be processed by industrial machines including, for example, industrial robots and service robots.

[0061] <Image Acquisition Unit 11b>

[0062] The image acquisition unit 11b acquires image data of a workpiece captured by a visual sensor 18 (such as a depth camera or stereo camera) or image data containing the three-dimensional shape of the workpiece, such as CAD data generated by a CAD (Computer Aided Design) system. The image acquisition unit 11b stores the acquired image data of the workpiece in the storage unit 16 (image DB 16a).

[0063] <Analysis Target Designation Unit 11c>

[0064] The analysis target designating unit 11 c acquires the content of the designation operation input by the user on the input screen for designating the analysis target displayed on the UI screen by the UI display control unit 11 a .

[0065] Figures 2 to 6 This is a schematic diagram showing an example of operations performed on the input screen for designating an analysis target.

[0066] exist Figure 2 In the example shown, as the user input for specifying the operation content, straight lines along the left and right sides of the image of the workpiece to be analyzed are input. Figure 2 In the case of the example shown, the feature extraction unit 11 d described later specifies that the user intends to analyze the width of the workpiece.

[0067] In addition, Figure 3 In the example shown, as the user input for specifying the operation content, a circle surrounding a hole formed in the workpiece in the image of the workpiece to be analyzed is input. Figure 3 In the case of the illustrated example, the feature extraction unit 11 d described later determines that the user intends to analyze the inner diameter of the hole formed in the workpiece.

[0068] In addition, Figure 4 In the example shown, as the user input for specifying the operation content, a circle surrounding the entire image of the workpiece to be analyzed is input. Figure 4 In the case of the example shown, for example, the user intends to perform centering (origin positioning) of a workpiece.

[0069] In addition, Figure 5 In the example shown, a rectangle that surrounds the entire image of the workpiece is input by specifying a diagonal line in the image of the workpiece to be analyzed. Figure 5 In the case of the example shown, the feature extraction unit 11 d described later determines that the user intends to perform centering (origin positioning) of the workpiece.

[0070] In addition, Figure 6 In the example shown, as the content of the designation operation input by the user, a point within the region of the workpiece in the image of the workpiece to be analyzed is designated. Figure 6 In the case of the example shown, the feature extraction unit 11 d described later determines that the user intends to perform centering (origin positioning) of the workpiece.

[0071] exist Figures 4 to 6 In the example shown, the feature extraction unit 11 d described later extracts the contour of the workpiece, and when the contour line is a two-dimensional quadrilateral, the center point of the workpiece is determined by measuring four sides and four vertices.

[0072] <Feature Extraction Unit 11d>

[0073] Based on the analysis target specification received by the analysis target specification unit 11c, the feature extraction unit 11d identifies the analysis target from the workpiece image and extracts (detects) the structural features (three-dimensional shape) of the identified analysis target. Furthermore, the feature extraction unit 11d converts the detected structural features into a real-world coordinate system (the worktable coordinate system) based on the distance image D of the RGBD image and the conversion formula between the display's plane coordinate system (the image coordinate system) and the worktable coordinate system.

[0074] Furthermore, the feature extraction unit 11 d stores the extracted structural features of the analysis target in the storage unit 16 (analysis target feature DB 16 b ).

[0075] In addition, these image coordinate systems and machine coordinate systems are calibrated and associated in advance. In this case, the coordinate system of the camera may be used instead of the plane coordinate system (image coordinate system) of the display.

[0076] Figures 7 to 11 1 is a schematic diagram showing an example of a process in which the feature extraction unit 11 d extracts a structural feature (three-dimensional shape) in accordance with a designation operation input by a user.

[0077] exist Figure 7 In the example shown, Figure 2 When the user inputs the left and right edges of the workpiece image as shown in FIG, the feature extraction unit 11d sets the vicinity of each segment (the input portion), extracts the contour line from the inner side thereof by the Canny method, and then detects two straight lines (the left and right outer edges of the workpiece) from the extracted contour line by the Hough transform. Figure 7 When a contour line is detected in this way, for example, if it is a contour line on the left, the feature extraction unit 11d can determine that there is an object on the right side of the contour line based on the fact that the pixels on the right side become convex values ​​when observed in the distance image D, or that there is an object on the right side due to background differences.

[0078] In addition, Figure 8 In the example shown, Figure 3 As shown, when the user inputs a circle surrounding a hole formed in the workpiece, the feature extraction unit 11d extracts a contour line from the inside of the circle surrounding the hole formed in the workpiece using a snake algorithm, and detects a circle (hole) based on the extracted contour line using Hough transform.

[0079] In addition, Figure 9 In the example shown, Figure 4 or Figure 5In a case where the user has input a circle or a rectangle that encloses the entire image of the workpiece as illustrated in FIG. 6A, the feature extraction section 11d extracts a contour line from the inside of the circle or the rectangle that encloses the entire image of the workpiece by a snake algorithm, detects four straight lines from the extracted contour line by a Hough transform, and detects a quadrangle (the upper, lower, left, and right outer edges of the workpiece) on the basis of the detected four straight lines.

[0080] Further, in a case where the user has input a circle or a rectangle that encloses the entire image of the workpiece as illustrated in FIG. 6A, the feature extraction section 11d extracts a contour line from the inside of the circle or the rectangle that encloses the entire image of the workpiece by a snake algorithm, detects four straight lines from the extracted contour line by a Hough transform, and detects a quadrangle (the upper, lower, left, and right outer edges of the workpiece) on the basis of the detected four straight lines. Figure 9 Figure 4 or Figure 5 In a case where the user has input a circle or a rectangle that encloses the entire image of the workpiece as illustrated in FIG. 6A, the feature extraction section 11d extracts a contour line from the inside of the circle or the rectangle that encloses the entire image of the workpiece by a snake algorithm, detects four straight lines from the extracted contour line by a Hough transform, and detects a quadrangle (the upper, lower, left, and right outer edges of the workpiece) on the basis of the detected four straight lines. Figure 10

[0081] Further, in a case where the user has input a circle or a rectangle that encloses the entire image of the workpiece as illustrated in FIG. 6A, the feature extraction section 11d extracts a contour line from the inside of the circle or the rectangle that encloses the entire image of the workpiece by a snake algorithm, detects four straight lines from the extracted contour line by a Hough transform, and detects a quadrangle (the upper, lower, left, and right outer edges of the workpiece) on the basis of the detected four straight lines. Figure 11 Figure 6 Further, in a case where the user has input a circle or a rectangle that encloses the entire image of the workpiece as illustrated in FIG. 6A, the feature extraction section 11d extracts a contour line from the inside of the circle or the rectangle that encloses the entire image of the workpiece by a snake algorithm, detects four straight lines from the extracted contour line by a Hough transform, and detects a quadrangle (the upper, lower, left, and right outer edges of the workpiece) on the basis of the detected four straight lines.

[0082] Further, in the above-described example, the structure extraction is performed by a 2-stage process of first performing the extraction of a contour by a snake algorithm or a canny method and then detecting a circle or a straight line by a Hough transform, but is not limited thereto. The extraction of a contour on the basis of a snake algorithm or a canny method is performed as a preprocessing for reducing false detection at the time of a Hough transform, and thus the extraction of a contour on the basis of a snake algorithm or a canny method can be skipped.

[0083] Figure 12 is a schematic view that indicates an example of conversion from an image coordinate system to a machine coordinate system.

[0084] As illustrated in FIG. 6B, the feature extraction section 11d converts the image coordinates of the detected structural feature of the analysis target (the left and right outer edges of the workpiece, a hole, the upper, lower, left, and right outer edges of the workpiece, and the like) into the coordinates of the real world, that is, the stage coordinates (machine coordinates). Figure 12

[0085] ​​​​This allows the feature extraction unit 11 d to acquire position information related to the three-dimensional shape of the entity to be analyzed designated by the user on the UI screen.

[0086] <Analysis Item Setting Unit 11e>

[0087] The analysis item setting unit 11 e receives designation of analysis items for the workpiece image.

[0088] For example, the analysis item setting unit 11 e sets analysis items for the structural features of the analysis target detected by the feature extraction unit 11 d by referring to a database (not shown) included in the storage unit 16 in which types of analysis items are defined.

[0089] Specifically, the analysis item setting unit 11 e lists items (analysis item candidates) that can be selected as analysis items based on the structural features of the analysis target detected by the feature extraction unit 11 d , referring to past analysis history information (analysis history DB 16 c ).

[0090] The analysis item setting unit 11e also sorts the listed analysis item candidates. Specifically, the analysis item setting unit 11e sorts the listed analysis item candidates based on the type, shape, processing status (before, during, and after processing) of the structure, the content of the processing program, and past analysis history. The analysis item candidates sorted by the analysis item setting unit 11e are displayed on the UI screen in sorted order by the UI display control unit 11a, and the user's selection of an analysis item is accepted.

[0091] Note that the operation of the analysis item setting unit 11e will be described later.

[0092] <Physical Quantity Calculation Unit 11f>

[0093] The physical quantity calculation unit 11 f extracts features required for calculating physical quantities related to the analysis items specified by the analysis item setting unit 11 e from the structural features of the analysis target detected by the feature extraction unit 11 d , and calculates physical quantities using the extracted features.

[0094] Hereinafter, the operation of the physical quantity calculation unit 11 f in each of the cases of (A) a cylindrical workpiece, (B) a rectangular parallelepiped workpiece with two holes formed therein, and (C) a cylindrical workpiece having a spherical surface shape will be described.

[0095] (A) Regarding cylindrical workpieces

[0096] like Figure 13As shown, the image acquisition unit 11b acquires, for example, image data of a cylindrical workpiece captured by the visual sensor 18 of a 3D camera, and the UI display control unit 11a displays the acquired RGB image on the UI screen. The analysis object designation unit 11c acquires the encirclement line for designating the analysis object input by the user via the input unit 14. Then, the feature extraction unit 11d uses the acquired encirclement line as the initial value and extracts the contour line of the outer side of the cylinder shown by the thick line through the snake algorithm. Next, the feature extraction unit 11d uses the contour line extracted by the snake algorithm and its inner side as the analysis object area, and extracts all the image features (contours) within the analysis object area. That is, as Figure 14 As shown, the feature extraction unit 11 d can extract two contour lines indicated by bold lines on the outside and inside of the cylindrical workpiece.

[0097] The feature extraction unit 11d converts each point on the extracted contour line into the real-world coordinate system (the table coordinate system) based on the distance image D of the RGBD image and the conversion formula between the image coordinate system and the table coordinate system. This allows accurate measurement of cylindrical workpieces that are distorted in the image and have proportions that differ from those in the real world. Furthermore, the feature extraction unit 11d performs circle and line detection on the contour converted to the table coordinate system, obtaining a structural feature such as the analysis target area consisting of two concentric circles.

[0098] Furthermore, the feature extraction unit 11d may also select structural features of the analysis target using additional information such as CAD data and machining programs. Alternatively, the user may specify the structural features on the UI screen.

[0099] Furthermore, the feature extraction unit 11 d can also acquire, from the range image D, a structural feature in which the analysis target region protrudes in the Z direction relative to the surroundings (the stage).

[0100] Next, when the feature extraction unit 11d detects the structural features of the cylindrical workpiece being analyzed, the analysis item setting unit 11e searches the analysis history DB 16c in the storage unit 16 using the structural features of the analysis target, namely, "workpiece machining condition," "concavity and convexity of the analysis target area," "contour segments," and "relationships between contour segments," as queries. Specifically, the analysis item setting unit 11e searches the analysis history DB 16c using the queries "workpiece machining condition: after machining" and the structural features of the cylindrical workpiece, namely, "concavity and convexity of the analysis target area: convex relative to surrounding area," "contour segments: circle × 2," and "relationships between contour segments: concentric." For example, the analysis item setting unit 11e lists selectable analysis items (candidate analysis items) such as "outer diameter," "height," "inner diameter," and "circumference."

[0101] The analysis item setting unit 11e then sorts the candidate analysis items listed in the analysis history DB 16c, which contains past analysis history records, by referring to the frequency of analysis items performed on the same cylindrical workpiece as the current one. Specifically, the analysis item setting unit 11e searches the analysis history DB 16c using the following queries: "Workpiece processing condition: After processing," "Concavity and convexity of analysis target area: Convex compared to surrounding area," "Contour segments: Circle × 2," and "Relationship between contour segments: Concentric." For example, if, in the past analysis history for the same cylindrical workpiece as the current one, the analysis results for "outer diameter" are "70%," "height" are "15%," "inner diameter" are "10%," and "circumference" are "5%," the analysis item setting unit 11e rearranges the candidate analysis items in the order of "outer diameter," "height," "inner diameter," and "circumference," indicating their highest likelihood of selection.

[0102] like Figure 15 As shown in FIG. 1 , the UI display control unit 11 a displays the analysis item candidates sorted by the analysis item setting unit 11 e on the UI screen in the sorted order, and then accepts the user's selection.

[0103] For example, when the user selects "outer diameter" as the desired analysis item from the candidate analysis items, the physical quantity calculation unit 11f extracts the outer circular contour line as a feature required for calculating the physical quantity of the analysis item in order to calculate the outer diameter. The physical quantity calculation unit 11f calculates the diameter of the circle using the coordinate values ​​of each point of the extracted circular contour line, for example, using a known method such as the least squares method. Figure 15 As shown, the UI display control unit 11 a may display the value of the diameter of the outer circle calculated by the physical quantity calculation unit 11 f on the UI screen.

[0104] Furthermore, the outer diameter of the cylindrical workpiece and the like may be calculated when the feature extraction unit 11 d converts each point on the contour line extracted from the cylindrical workpiece into a table coordinate system.

[0105] (B) Regarding a rectangular workpiece with two holes

[0106] like Figure 16 As shown, the image acquisition unit 11 b acquires image data of a rectangular parallelepiped workpiece having two holes formed therein, which is captured from obliquely above by the visual sensor 18 of the 3D camera.

[0107] The UI display control unit 11a displays the acquired RGB image on the UI screen. Figure 6Similarly to the case of , the analysis target designation unit 11c obtains the point for designating the analysis target input by the user via the input unit 14. Then, the feature extraction unit 11d obtains the area having the same brightness value as the vicinity of the obtained point by the region expansion method, and sets the area indicated by the hatching as the analysis target area. Figure 17 As shown in FIG. 1 , the feature extraction unit 11 d extracts all the image features (contours) indicated by bold lines within the analysis target region.

[0108] Based on the distance image D of the RGBD image and the conversion formula between the image coordinate system and the worktable coordinate system, the feature extraction unit 11d converts each point on the extracted contour line into the real-world coordinate system (the worktable coordinate system). This allows accurate measurement of a rectangular workpiece that is distorted in the image and has two holes with proportions different from those in the real world. The feature extraction unit 11d then performs circle detection, line detection, and cuboid detection on the contour converted to the worktable coordinate system, obtaining a structural feature that shows the analysis target area is composed of nine contour lines and two circles that form a portion of the cuboid. Furthermore, the feature extraction unit 11d can also obtain a structural feature from the distance image D that shows the analysis target area protruding in the Z direction relative to its surroundings (the worktable).

[0109] Next, when the feature extraction unit 11d detects the structural features of the rectangular workpiece with two holes as the target for analysis, the analysis item setting unit 11e searches the analysis history DB 16c in the storage unit 16 using the structural features of the analysis target, namely, "workpiece machining condition," "concavity and convexity of the analysis target area," "contour segments," and "relationships between contour segments," as queries. Specifically, the analysis item setting unit 11e searches the analysis history DB 16c using the queries "workpiece machining condition: after machining" and the structural features of the rectangular workpiece with two holes, namely, "concavity and convexity of the analysis target area: convex relative to surrounding area," "contour segments: 9 straight lines and 2 circles," and "relationships between contour segments: rectangular." For example, the analysis item setting unit 11e lists items (candidate analysis items) that can be selected as analysis items, such as "center of rectangular workpiece," "height of rectangular workpiece," and "inner diameter of circle."

[0110] The analysis item setting unit 11e then sorts the listed analysis item candidates in the analysis history DB 16c, which represents past analysis history, by referring to the frequency of analysis items performed on the same workpiece as the current rectangular workpiece with two holes. Specifically, the analysis item setting unit 11e searches the analysis history DB 16c using the queries "Workpiece processing status: After processing," "Concavity and convexity of analysis target area: Convex compared to surrounding area," "Contour segments: 9 straight lines and 2 circles," and "Relationship between contour segments: Rectangular." For example, if, in the past analysis history for the same analysis target as the current rectangular workpiece with two holes, the analysis results for "Cuboid center" are "60%," "Cuboid height" are "25%," and "Circle inner diameter" are "5%," the analysis item setting unit 11e rearranges the analysis item candidates in the order of "Cuboid center," "Cuboid height," and "Circle inner diameter," which are most likely to be selected.

[0111] like Figure 18 As shown, the UI display control unit 11a displays the candidate analysis items sorted by the analysis item setting unit 11e on the UI screen in the sorted order. The user's selection is then accepted. Alternatively, the top-ranked candidate analysis item can be automatically set as the analysis item, or the Nth-ranked analysis item can be set in advance (N is an integer greater than or equal to 2).

[0112] For example, when the user selects "center of a cuboid" as the desired analysis item from the candidate analysis items, the physical quantity calculation unit 11f extracts 9 straight contour lines as features required for calculating the physical quantity of the analysis item in order to calculate the center coordinates of the cuboid. The physical quantity calculation unit 11f uses 4 of the extracted straight contour lines to calculate the midpoints of the three faces of the cuboid. The physical quantity calculation unit 11f calculates the center coordinates of the cuboid based on the calculated midpoints of the three faces of the cuboid. Figure 18 As shown, the UI display control unit 11 a may display the center coordinates of the cuboid calculated by the physical quantity calculation unit 11 f on the UI screen.

[0113] Furthermore, the feature extraction unit 11 d may calculate the center coordinates of the rectangular parallelepiped workpiece having two holes when converting each point on the contour line extracted from the rectangular parallelepiped workpiece having two holes into the table coordinate system.

[0114] (C) Regarding a cylindrical workpiece having a spherical shape

[0115] like Figure 19As shown, the image acquisition unit 11b acquires image data of a cylindrical workpiece having a spherical surface, captured from directly above by the 3D camera's visual sensor 18. The UI display control unit 11a displays the acquired RGB image on the UI screen. The analysis target designation unit 11c acquires a bounding line specifying the analysis target, input by the user via the input unit 14. The feature extraction unit 11d then uses the acquired bounding line as an initial value to extract the contour of the hemispherical portion indicated by the bold line using a snake-shaped algorithm. The feature extraction unit 11d then uses the contour line extracted by the snake-shaped algorithm and its inner portion as the analysis target area and extracts image features (contours and image coordinate values ​​of each pixel) within the analysis target area.

[0116] The feature extraction unit 11d converts the extracted image coordinate values ​​of each pixel into a real-world coordinate system (the stage coordinate system) based on the distance image D of the RGBD image and the conversion formula between the image coordinate system and the stage coordinate system. The feature extraction unit 11d then performs circle detection, etc. on the contour lines converted to the stage coordinate system, to obtain a structural feature indicating that the analysis target area consists of a single circle (i.e., a single continuous surface).

[0117] Furthermore, the feature extraction unit 11d can perform sphere detection on the extracted point group (each pixel converted to the worktable coordinate system) to obtain a structural feature that the analysis target area is a sphere. Furthermore, the feature extraction unit 11d can also obtain a structural feature from the range image D that the analysis target area protrudes in the Z direction relative to the surrounding area (the worktable).

[0118] Furthermore, the feature extraction unit 11d may also select structural features of the analysis target using additional information such as CAD data and machining programs. Alternatively, the user may specify the structural features on the UI screen.

[0119] Next, when the feature extraction unit 11d detects the structural features of a cylindrical workpiece with a spherical surface as the target for analysis, the analysis item setting unit 11e searches the analysis history DB 16c in the storage unit 16 using the workpiece's structural features, including "workpiece machining conditions," "concavity and convexity of the analysis target area," "surface segments," and "relationships between surface segments," as queries. Specifically, the analysis item setting unit 11e searches the analysis history DB 16c using the queries "workpiece machining conditions: after machining" and the structural features of a cylindrical workpiece with a spherical surface, including "concavity and convexity of the analysis target area: convex compared to surrounding area," "surface segments: 3D spherical surface × 1," and "relationships between surface segments: unknown (-)." For example, the analysis item setting unit 11e lists items (candidate analysis items) that can be selected as analysis items, such as "surface roughness," "sphere center," and "sphere diameter."

[0120] The analysis item setting unit 11e then sorts the candidate analysis items listed in the analysis history DB 16c, which contains past analysis history records, by referring to the frequency of analysis items performed on the same cylindrical workpiece with a spherical surface as the current analysis item. Specifically, the analysis item setting unit 11e searches the analysis history DB 16c using the following queries: "Workpiece processing status: After processing," "Concavity and convexity of analysis target area: Convex compared to surrounding area," "Surface segment: 3D spherical surface × 1," and "Relationship between surface segments: Unknown (-)." For example, if, in the past analysis history for the same cylindrical workpiece with a spherical surface as the current analysis item, the analysis item setting unit 11e finds that the analysis results for "surface roughness" are "50%," "sphere center" are "25%," and "sphere diameter" are "25%," the analysis item setting unit 11e rearranges the candidate analysis items in the order of "surface roughness," "sphere center," and "sphere diameter," which are most likely to be selected.

[0121] like Figure 21 As shown in FIG. 1 , the UI display control unit 11 a displays the analysis item candidates sorted by the analysis item setting unit 11 e on the UI screen in the sorted order, and then accepts the user's selection.

[0122] Alternatively, all candidates for analysis items may be automatically set as analysis items. In this case, it is preferable to determine in advance that all candidates for analysis items are set as analysis items.

[0123] For example, when all candidates for analysis items are set as analysis items, the physical quantity calculation unit 11f extracts the table coordinate values ​​of the point group (each pixel converted into the table coordinates) in the analysis object area in order to calculate the surface roughness of the hemisphere, etc. The physical quantity calculation unit 11f uses the table coordinate values ​​of the extracted point group to calculate the parameters of the sphere (the center and diameter of the sphere) according to a well-known method such as the least squares method. At this time, the physical quantity calculation unit 11f also calculates the square root of the square error (RMS or Rq) as the "surface roughness". As Figure 21 As shown, the UI display control unit 11 a may display Rq (surface roughness), the spherical center, and the spherical diameter calculated by the physical quantity calculation unit 11 f on the UI screen.

[0124] Alternatively, the parameters of the spherical surface (center and diameter of the sphere) may be calculated by converting the image coordinate values ​​of each pixel extracted by the feature extraction unit 11d into the stage coordinate system. Alternatively, the parameters of the spherical surface (center and diameter of the sphere) may be calculated using information from CAD data.

[0125] Furthermore, while the above embodiments describe cylindrical workpieces with spherical surfaces, the shape is not limited to this. For example, a workpiece with a free-form surface may also be used. In this case, after extracting a point cloud using the same steps as for spherical workpieces, physical quantities such as surface roughness can be calculated, for example, by comparing the point cloud with information from CAD data.

[0126] <Image Analysis Processing by Workpiece Image Analysis Device 1>

[0127] Next, refer to Figure 22 The processing flow of the workpiece image analysis device 1 will be described.

[0128] Figure 22 This is a flowchart for explaining the image analysis processing of the workpiece image analysis device 1 .

[0129] The image analysis process is started by inputting an instruction to start the image analysis process via the input unit 14 .

[0130] In step S1 , the UI display control unit 11 a displays a user interface screen (UI screen) for the user to input and output various information during the image analysis process.

[0131] In step S2, the image acquisition unit 11b acquires image data of the workpiece captured by the visual sensor 18 (such as a depth camera or stereo camera) or image data containing the three-dimensional shape of the workpiece, such as CAD data generated by a CAD (Computer Aided Design) system. The acquired workpiece image data is stored in the storage unit 16.

[0132] In step S3 , the UI display control unit 11 a displays an input screen for designating an analysis target in the acquired workpiece image.

[0133] In step S4 , the analysis target designating unit 11 c acquires the details of the designation operation input by the user on the input screen for designating the analysis target.

[0134] In step S5 , the feature extraction unit 11 d specifies an analysis target region, which is a portion to be analyzed, in the image of the workpiece based on the content of the designation operation input by the user.

[0135] In step S6 , the feature extraction unit 11 d extracts image features (contours) of the identified analysis target region.

[0136] In step S7 , the feature extraction unit 11 d converts the extracted image features (contours) from the plane coordinate system of the display (image coordinate system) to the real-world coordinate system (table coordinate system) on the table on which the workpiece is placed.

[0137] In step S8 , the feature extraction unit 11 d performs circle detection, line detection, and the like on the image features (contours) converted into the real-world coordinate system, thereby acquiring structural features of the analysis target.

[0138] In step S9 , the analysis item setting unit 11 e lists candidates of analysis items that can be selected as analysis items by referring to past analysis records (analysis record DB 16 c of the storage unit 16 ) based on the acquired structural features of the analysis target.

[0139] In step S10 , the analysis item setting unit 11 e ranks the listed analysis item candidates based on past analysis history and the like.

[0140] In step S11 , the UI display control unit 11 a displays the analysis item candidates sorted by the analysis item setting unit 11 e on the UI screen in the sorted order.

[0141] In step S12 , the UI display control unit 11 a displays an input screen for accepting selection from among analysis item candidates, and accepts the user's selection.

[0142] In step S13 , the physical quantity calculation unit 11 f extracts features (profiles) necessary for calculating the physical quantity of the selected analysis item from the data of the analysis target region converted into the stage coordinate system.

[0143] In step S14 , the physical quantity calculation unit 11 f calculates the physical quantity of the selected analysis item based on the extracted feature (profile).

[0144] In step S15 , the UI display control unit 11 a displays the physical quantity calculated in step S14 on the UI screen.

[0145] After step S15 , the image analysis process ends.

[0146] As described above, the workpiece image analysis device 1 according to one embodiment accepts a user operation to specify an analysis target in an image of a workpiece. Based on the user's designation, the workpiece image analysis device 1 identifies an analysis target region in the workpiece image and extracts the structural features of the analysis target in a real-world coordinate system. The workpiece image analysis device 1 displays candidate analysis items that can be selected from the extracted structural features of the analysis target on a user interface screen and calculates the physical quantity of the analysis item selected by the user.

[0147] As a result, the workpiece image analysis device 1 eliminates the need for complex steps and specialized knowledge related to image analysis settings, as well as measuring instruments such as contact probes, required for calculating predetermined physical quantities related to analysis items. Instead, it can perform high-speed analysis using only the visual sensor 18. Furthermore, the workpiece image analysis device 1 can reduce the workload required for workpiece measurement.

[0148] Although one embodiment has been described above, the workpiece image analysis device 1 is not limited to the above-described embodiment, and includes modifications and improvements within a range that can achieve the purpose.

[0149] <Variation 1>

[0150] In the above embodiment, the workpiece image analysis device 1 uses a single image captured by the vision sensor 18 to calculate the physical quantity of the analysis item selected by the user, but this is not limited to this. For example, the workpiece image analysis device 1 may move the vision sensor 18 relative to the workpiece, acquire multiple images of the workpiece, and generate a single high-resolution image of all of the workpieces.

[0151] Thus, the workpiece image analysis device 1 can acquire an image of the entire workpiece even if the workpiece is large and does not fit within the imaging range of one visual sensor.

[0152] In addition to the relative movement of the vision sensors, multiple images can also be acquired by capturing images with multiple vision sensors.

[0153] At this time, regarding the designation of the analysis target, for example, the analysis target may be designated for each of the plurality of images in the designation of the analysis target in step S4 , and the features extracted from the analysis targets in step S7 may be integrated.

[0154] <Variation 1-2>

[0155] For example, in the above embodiment, the vision sensor 18 may be moved relatively based on the resolution setting information of the numerical control device or the machine tool and then captured so as to obtain the required resolution when calculating the physical quantity.

[0156] For example, consider a case where a numerical controller is set to control a machine tool with a positioning resolution of 1 μm.

[0157] At this time, for example, when the user specifies a rectangular workpiece as the analysis object and the analysis item is "center of gravity position", the physical quantity finally calculated by the physical quantity calculation unit requires a resolution of several μm or less in order to use the physical quantity in the machine tool.

[0158] Typically, in image measurement using a vision sensor, the target resolution of the vision sensor is expressed as the ratio of the number of pixels to the image sensor's field of view. For example, if the image sensor's field of view is a 10 cm square and the camera has 1000 pixels in height x 1000 in width, the target resolution is 100 μm.

[0159] The resolution of the vision sensor can be improved by, for example, bringing the relative positions of the vision sensor and the workpiece closer and narrowing the imaging range of the vision sensor.

[0160] Therefore, for example, if the resolution setting information for the numerical controller is set to the target resolution, the analysis target can be specified by capturing a wide-area image, such as one that includes the entire workpiece. Before extracting the features of the analysis target in step S6, the vision sensor can be relatively moved to achieve the target resolution, and then the image can be captured again. Furthermore, if the target resolution cannot be achieved despite relative movement of the vision sensor, a warning message can be displayed, for example.

[0161] This makes it possible to obtain the resolution required for calculating physical quantities.

[0162] <Variation 2>

[0163] Furthermore, for example, in the above-described embodiment, the workpiece image analysis apparatus 1 allows the user to select an analysis item after specifying the analysis target and acquiring the structural features of the analysis target. However, the present invention is not limited to this. For example, the workpiece image analysis apparatus 1 may allow the user to select an analysis item, then specify the analysis target and acquire the structural features of the analysis target.

[0164] Thus, the workpiece image analysis device 1 limits the analysis target by selecting the analysis item, and thus can more appropriately detect the structural features that become the analysis target.

[0165] <Variation 3>

[0166] Furthermore, for example, in the above embodiment, the workpiece image analysis apparatus 1 lists the analysis item candidates and then sorts the analysis item candidates based on the analysis history DB 16c, but the present invention is not limited thereto. For example, the workpiece image analysis apparatus 1 may simply list all analysis items without sorting.

[0167] Furthermore, the workpiece image analysis device 1 may list analysis items based not only on the analysis history DB 16 c but also on an estimator using machine learning.

[0168] <Variation 4>

[0169] Furthermore, for example, in the above-described embodiment, the visual sensor 18 is a depth camera or a 3D camera, but this is not limiting. For example, if the shape of the workpiece is known in advance and the height of the workpiece is fixed, the visual sensor 18 may also be a 2D camera. The following description assumes that the visual sensor 18 is a 2D camera.

[0170] For example, in Figure 13 In the case of a cylindrical workpiece, the visual sensor 18 captures the cylindrical workpiece from directly above. The image acquisition unit 11b acquires the image data of the cylindrical workpiece captured by the visual sensor 18 from directly above, and the UI display control unit 11a displays the acquired image on the UI screen. Figure 13 Similarly to the case of , the analysis target designation unit 11c obtains the encircling line of the analysis target input by the designated user. Then, the feature extraction unit 11d uses the obtained encircling line as the initial value and extracts the contour line of the outer side of the cylinder shown by the thick line using the snake algorithm. Next, the feature extraction unit 11d uses the contour line extracted by the snake algorithm and its inner side as the analysis target area and extracts all the image features (contours) within the analysis target area. In this way, Figure 14 Similarly, the feature extraction unit 11d can extract the two contour lines (shown as bold lines) on the outer and inner sides of the cylindrical workpiece. Furthermore, the feature extraction unit 11d can also use additional information such as CAD data and machining programs to select structural features of the analysis target. Alternatively, the user can specify the feature on the display.

[0171] In addition, the feature extraction unit 11d converts each point on the extracted contour line into the coordinate system of the real world (the worktable coordinate system) based on the distance between the visual sensor 18 and the cylindrical workpiece calculated in advance by the focusing method and the conversion formula between the image coordinate system and the worktable coordinate system, so as to accurately measure the cylindrical workpiece that is distorted in the image and has a different proportion from the real world. Figure 14 Similarly to the case of , the feature extraction unit 11d can perform circle detection, line detection, etc. on the contour converted into the stage coordinate system, and obtain a structural feature that the analysis target area is composed of two concentric circles.

[0172] Furthermore, because the previously acquired brightness value of the workbench is identical to the brightness value surrounding the target analysis area, the feature extraction unit 11d can determine that the area surrounding the target analysis area is a workbench with a height of "0." Therefore, the feature extraction unit 11d can also extract a structural feature that the target analysis area protrudes in the Z direction relative to the surrounding area (the workbench). Subsequent processing is the same as that performed when the visual sensor 18 is a depth camera or a 3D camera, and therefore, further explanation is omitted.

[0173] In addition, when mass-producing the same workpiece, the analysis object area, features, physical quantity calculation steps, information on the relative movement of the camera, or the workpiece image used for the first workpiece can also be saved in advance. In the second and subsequent workpieces, the position and angle movement amount relative to the first workpiece image is calculated, and the coordinate values ​​of the analysis object area and features generated for the first workpiece are added with the said movement amount, and the physical quantity calculation steps are corrected, thereby skipping the user's designation operation of the analysis object and decision operation of the analysis item for the second and subsequent workpieces, making the analysis action completely automated.

[0174] Furthermore, each function included in the workpiece image analysis device 1 of one embodiment can be realized by hardware, software, or a combination thereof. Here, realization by software means realization by having a computer read a program and execute it.

[0175] In addition, each component included in the workpiece image analysis device 1 can be realized by hardware including electronic circuits, software, or a combination thereof.

[0176] Any type of non-transitory computer readable medium (Non-transitory computerreadable medium) can be used to store the program and provide it to the computer. Non-transitory computer readable media include various types of tangible recording media (Tangible storage medium). Examples of non-transitory computer readable media include magnetic recording media (e.g., floppy disks, magnetic tapes, hard disk drives, etc.), optical magnetic recording media (e.g., magneto-optical disks), CD-ROMs (Read Only Memory), CD-Rs, CD-R / Ws, and semiconductor memories (e.g., mask ROMs, PROMs (Programmable ROMs), EPROMs (Erasable PROMs), flash ROMs, RAMs, etc.). In addition, the program can also be provided to the computer via various types of transient computer readable media (Transitory computerreadable medium). Examples of transient computer readable media include electric signals, optical signals, and electromagnetic waves. Transitory computer readable media can provide the program to the computer via wired communication lines such as electric wires and optical fibers, or wireless communication lines.

[0177] Furthermore, the steps describing the program recorded in the recording medium naturally include processing that is performed in time series according to the order, but also include processing that is not necessarily performed in time series and is executed in parallel or individually.

[0178] In other words, the workpiece image analysis device, workpiece image analysis method, and program of the present disclosure can adopt various embodiments having the following structures.

[0179] (1) The workpiece image analysis device 1 disclosed in the present invention comprises: a display unit 15, which displays an image of a workpiece to be processed by an industrial machine; an analysis object specifying unit 11c, which receives a specification of an analysis object for the image of the workpiece; a feature extraction unit 11d, which extracts structural features of the analysis object specified by the analysis object specifying unit 11c in a real-world coordinate system; an analysis item setting unit 11e, which receives a specification of an analysis item for the image of the workpiece; and a physical quantity calculation unit 11f, which extracts features required for calculating physical quantities involved in the analysis items specified by the analysis item setting unit 11e from the structural features of the analysis object, and calculates the physical quantities using the extracted features.

[0180] According to the workpiece image analysis device 1 , the workload required for measuring a workpiece can be reduced.

[0181] (2) In the workpiece image analysis device 1 described in (1), the analysis item setting unit 11 e ranks and displays candidates for analysis items for the structural features of the analysis target.

[0182] Thereby, the workpiece image analysis apparatus 1 can reliably designate analysis items desired by the user.

[0183] (3) In the workpiece image analysis device 1 described in (1) or (2), the workpiece image may be at least one of a two-dimensional or three-dimensional captured image of the workpiece and an image of CAD data of the workpiece.

[0184] Thereby, the workpiece image analysis apparatus 1 can accurately acquire the structural features of the analysis target.

[0185] (4) The workpiece image analysis method disclosed in the present invention comprises: a display step for displaying an image of a workpiece that is a processing object of an industrial machine; an analysis object designation step for receiving a designation of an analysis object for the image of the workpiece; a feature extraction step for extracting structural features of the analysis object designated in the analysis object designation step in a real-world coordinate system; an analysis item setting step for receiving a designation of an analysis item for the image of the workpiece; and a physical quantity calculation step for extracting features required for calculating physical quantities involved in the analysis items designated in the analysis item setting step from the structural features of the analysis object, and using the extracted features to calculate the physical quantities.

[0186] According to this workpiece image analysis method, the same effect as (1) can be achieved.

[0187] (5) The program disclosed in the present invention enables a computer to implement the following functions: a display control function that displays an image of a workpiece that is a processing object of an industrial machine; an analysis object designation function that receives a designation of an analysis object for the image of the workpiece; a feature extraction function that extracts structural features of the analysis object specified in the analysis object designation function in a real-world coordinate system; an analysis item setting function that receives a designation of an analysis item for the image of the workpiece; and a physical quantity calculation function that extracts features required for calculating physical quantities involved in the analysis item specified in the analysis item setting function from the structural features of the analysis object, and uses the extracted features to calculate the physical quantities.

[0188] According to this program, the same effect as (1) can be achieved.

[0189] Explanation of symbols

[0190] 1 Workpiece image analysis device

[0191] 11 CPU

[0192] 11a UI display control unit

[0193] 11b Image acquisition unit

[0194] 11c Analysis target specification section

[0195] 11d Feature Extraction Unit

[0196] 11e Analysis Project Setting Department

[0197] 11f Physical Quantity Calculation Department

[0198] 12 ROM

[0199] 13 RAM

[0200] 14 Input section

[0201] 15 Display

[0202] 16 Storage

[0203] 16a Image Database

[0204] 16b Parsing Object Feature Database

[0205] 16c Parsing Resume Database

[0206] 17 Ministry of Communications

[0207] 18 vision sensors.

Claims

1. A workpiece image analysis device, characterized in that: have: a display unit that displays an image of a workpiece to be processed by the industrial machine; an analysis target designating unit that receives a designation of an analysis target, the analysis target including any one of a line representing at least a portion of the outer shape of the workpiece in the image of the workpiece, an internal region including at least a portion of a region of the workpiece, and at least one point within the region of the workpiece; a feature extraction unit that determines the analysis target specified by the analysis target specifying unit based on the image of the workpiece, detects structural features of the analysis target, and extracts the detected structural features in a real-world coordinate system based on the range image of the workpiece and a conversion formula between the image coordinate system and the worktable coordinate system; an analysis item setting unit that receives a designation of an analysis item for the detected structural feature of the analysis target; as well as The physical quantity calculation unit extracts features required for calculation of a physical quantity related to the analysis item specified by the analysis item setting unit from the structural features of the analysis object, and calculates the physical quantity using the extracted features.

2. The workpiece image analysis device according to claim 1, wherein The analysis item setting unit ranks and displays candidates for analysis items corresponding to the structural features of the analysis target.

3. The workpiece image analysis device according to claim 1 or 2, characterized in that: The image of the workpiece is at least one of a two-dimensional or three-dimensional captured image of the workpiece and an image of CAD data of the workpiece.

4. A workpiece image analysis method, characterized in that: The following steps are involved: a display step of displaying an image of a workpiece to be processed by the industrial machine; an analysis target designation step of receiving a designation of an analysis target, the analysis target including any one of a line representing at least a portion of the outer shape of the workpiece in the image of the workpiece, an internal region including at least a portion of a region of the workpiece, and at least one point within the region of the workpiece; a feature extraction step of determining the analysis object specified in the analysis object specifying step based on the image of the workpiece, detecting structural features of the analysis object, and extracting the detected structural features in a real-world coordinate system based on the range image of the workpiece and a conversion formula between the image coordinate system and the worktable coordinate system; an analysis item setting step of receiving a designation of an analysis item for the detected structural feature of the analysis object; as well as The physical quantity calculation step extracts features required for calculating the physical quantity related to the analysis item specified in the analysis item setting step from the structural features of the analysis object, and calculates the physical quantity using the extracted features.

5. A computer-readable medium having a program recorded thereon, characterized in that: The program enables the computer to implement the following functions: a display control function for displaying an image of a workpiece to be processed by the industrial machine; an analysis object designation function that receives a designation of an analysis object, the analysis object including any one of a line representing at least a portion of the outer shape of the workpiece in the image of the workpiece, an internal region including at least a portion of a region of the workpiece, and at least one point within the region of the workpiece; a feature extraction function for determining the analysis object specified in the analysis object specification function based on the image of the workpiece, detecting structural features of the analysis object, and extracting the detected structural features in a real-world coordinate system based on the range image of the workpiece and a conversion formula between the image coordinate system and the worktable coordinate system; an analysis item setting function for receiving a designation of an analysis item for the detected structural feature of the analysis object; as well as A physical quantity calculation function extracts features required for calculating a physical quantity related to an analysis item specified in the analysis item setting function from structural features of the analysis object, and calculates the physical quantity using the extracted features.

Images