Visual identification and profile feature measurement system and method for irregular cracks of medium-thickness plate

Abstract

The invention belongs to the field of intelligent measurement of metallurgical equipment, and discloses a system and a method for visually identifying irregular cracks of a medium-thickness plate and measuring profile features, and the system comprises a visual system consisting of an optical machine and a camera, a movable portal frame for installing the measurement system, and a medium-thickness plate irregular crack calculation system, after the steel plate is stopped, the medium-thickness plate is divided into a plurality of areas according to the size of the medium-thickness plate and the minimum coverage area of the visual system, the moving portal frame moves to the head position of the medium-thickness plate, and the measuring system starts measuring the first area of the medium-thickness plate and then moves one area by one area until measurement is completed. According to the invention, discontinuous cracks of the medium-thickness plate can be accurately found through a grating image processing method, the spatial position, length and area of the cracks can be measured, the system does not need to process areas outside the cracks, and rapid and accurate measurement of irregular cracks is ensured.

Description

TECHNICAL FIELD

[0001] The present application belongs to the field of intelligent measurement of metallurgical equipment, and particularly relates to a system and method for visual identification and shape feature measurement of irregular cracks of a medium plate. BACKGROUND

[0002] A medium plate is a common material used in the fields of construction, manufacturing and shipbuilding. In these applications, the strength of the medium plate is crucial to the quality of the product. Irregular cracks are one of the important factors that restrict the strength of the medium plate. At present, the commonly used method for detecting irregular cracks and measuring the spatial position, length and area shape feature of the medium plate includes using ultrasonic detection and manual observation. This method is manually operated and has low measurement efficiency. The use of machine vision technology to detect irregular cracks and measure the shape feature of the medium plate can improve the measurement speed and accuracy. Therefore, the irregular crack identification and shape feature measurement method based on machine vision is necessary.

[0003] At present, the image recognition method is the main means for detecting defects of the medium plate. This method can accurately identify the defects of the medium plate. Since cracks directly affect the strength, hardness and toughness of the medium plate, it is necessary to accurately identify irregular cracks and measure the length and area of the cracks for repair. The image processing method can only identify that there is a crack at the position and cannot measure the size of the irregular crack, which seriously affects the pre-detection effect of the medium plate and the subsequent repair process of the medium plate. SUMMARY

[0004] To solve the problems existing in the prior art, the present application provides a system and method for visual identification and shape feature measurement of irregular cracks of a medium plate. The method of grating image processing can accurately find the discontinuous cracks of the medium plate and measure the spatial position, length and area of the cracks. The system does not need to process the areas other than the cracks, ensuring the rapid and accurate measurement of irregular cracks.

[0005] To achieve the above-mentioned purpose, the present application provides the following solutions. A system for visual identification and shape feature measurement of irregular cracks of a medium plate, comprising: a vision system composed of a light machine and a camera, a mobile gantry mounting the vision system, and a medium plate irregular crack calculation system. After the steel plate stops, the medium plate is divided into multiple areas according to the size of the medium plate and the minimum coverage area of the vision system. The mobile gantry moves to the head position of the medium plate. The measurement system starts to measure from the first area of the medium plate, and then moves area by area until the measurement is completed. The light machine in the measurement system projects the modulated superimposed composite grating to the surface of the plate once, and the composite pattern modulated by the plate surface is captured by the external trigger camera; the discontinuous position in the phase map solved by capturing the composite grating image is segmented to obtain the discontinuous coordinate position of the irregular crack in the plate in the absolute phase map, and the discontinuous position is mapped to the point cloud to calculate the length and area of the irregular crack of the plate.

[0006] Preferably, the visual system composed of the light machine and the camera includes: The light machine with a burnable self-encoding projection image is set to an external trigger projection mode, the self-encoding projection image is a computer-generated composite grating image, the composite grating image is burned into the memory of the light machine through the light machine burning software, and the projection interval time and the projection mode of the light machine are set; The camera is an industrial color camera directly connected to the external trigger of the light machine, which triggers the camera to capture images and sends them to the plate irregular crack calculation system after the light machine projects the images.

[0007] Preferably, the generation process of the projected composite grating image is: For the projected composite grating image, each individual grating image is modulated with a carrier image and superimposed together, and the projected composite grating image accumulates each modulated image, i.e. ; In the formula, is the projected grating fringe image; is the carrier frequency of the projected grating fringe image; is the transverse coordinate of the projected grating fringe image; is the phase of the three-step phase shift, ; and are the background intensity and the reflection intensity, respectively; is the number of projected grating fringe images.

[0008] Preferably, the mobile gantry on which the visual system is installed includes: A metal switch is installed below the roller, which triggers the metal switch to control the roller to stop moving when the steel plate is transported below the roller, the steel plate is regionally segmented according to the size information of the steel plate, the size of the segmented region is the projection region of the light machine, then the gantry is controlled to move to the head position of the steel plate, taking the upper left corner of the steel plate as the starting point of projection, and the measurement is performed region by region in the order from left to right until the measurement is completed.

[0009] Preferably, the plate irregular crack calculation system includes: The image captured by the camera is uploaded to the server, the single composite grating modulated by the surface of the steel plate captured by the camera is phase solved in the server, the discontinuous positions are segmented in the phase image, the irregular crack area is obtained by mapping the discontinuous positions to the three-dimensional point cloud, and the visualization display is performed on the upper computer.

[0010] Preferably, the phase solving process of the composite grating image captured by the camera includes: The captured image is recovered by a band-pass and low-pass filter to recover the pattern of each modulation channel, and the demodulated pattern is a single grating image used for calculating the wrapped phase, which is represented as: ; In the formula, is the average gray scale of the grating image; is the modulation gray scale of the pixel; is the wrapped phase to be solved; is the current phase shift; is the pixel coordinate; The solvable phase of the obtained single grating image is: ; Wherein, is the demodulated single grating image; n is the initial phase of the single grating image. n

[0011] Preferably, the solved phase is unfolded into absolute phase by spatial phase unfolding, and the process of obtaining the discontinuous position in the absolute phase includes: The absolute phase of the matching point is solved as: ; Wherein, the integer multiple is the wrapping number; The least square direction of the unfolded absolute phase image is estimated as: ; If there is a discontinuous position in the phase image, two different direction regions are represented by two unequal eigenvalues of the phase image, and are the eigenvectors corresponding to the two eigenvalues respectively; the bilateral filter is applied to the least square direction image, and the edges / texture of the bilateral filter output are extracted / removed respectively, and the output image is clearly marked as two regions: the continuous region with a value of 0 and the discontinuous region with a value of 1; the weight mask is obtained by re-marking the two regions, that is, the weight mask is 0 corresponding to the phase continuous region, and the weight mask is 1 corresponding to the phase discontinuous region. ​​​

[0012] Preferably, based on the discontinuities corresponding to the irregular crack locations in the calculated phase, establishing the mapping relationship between the discontinuous phase and the three-dimensional point cloud of the crack includes: ; in, These are the three-dimensional point cloud coordinates of the crack; For mapping functions; The absolute phase of the matching point; For the rows of patterns projected by the projector; The frequency of the stripes in the projected pattern.

[0013] Preferably, based on the solved 3D point cloud information of the irregular crack, the calculation process of the surface features of the irregular crack is as follows: The spatial location of an irregular crack is represented by its geometric center and direction vector. The geometric center represents the spatial location of the crack, and the direction vector represents the direction of the irregular crack. for: ; in, These represent the first and second irregular cracks. i Three-dimensional coordinate information of each point, Indicates the total number of irregular cracks n One point; For the first irregular crack i There are points, and the direction vector of each point is: ; Based on the obtained 3D point cloud information of the irregular crack, the process for determining the length of the irregular crack is as follows: The 3D point cloud representation of irregular cracks is as follows , For the first irregular crack n The length of the irregular crack is calculated based on the three-dimensional coordinate information of each point and the cumulative distance of the path points. L for: ; Based on the obtained 3D point cloud information of the irregular crack, the process for calculating the area of ​​the irregular crack is as follows: Calculating the area enclosed by the 3D point cloud of an irregular crack requires reducing the 3D point cloud to a 2D plane using PCA projection. The PCA projection process includes data centralization, covariance matrix calculation, and PCA decomposition, as detailed below: ; in, For the centralized point cloud, The number of point clouds; covariance matrix for: ; in, for The transpose of the matrix; For covariance matrix The eigenvalue decomposition process is as follows: ; in, It is the characteristic matrix; Select the first two principal components Using a 2D planar coordinate system as the basis, projecting 3D points onto this plane is represented as: ; For the projected 2D point Calculate the area enclosed by the points on the outer contour of the irregular crack using the following formula. ,Right now: .

[0014] This invention also provides a method for visual recognition and surface feature measurement of irregular cracks in medium-thick plates. The method is implemented using the aforementioned system and includes: Step 1: Obtain the composite grating image; Step 2: Set the optical engine with programmable self-encoded projection images to external trigger projection mode. The self-encoded projection image is a computer-generated composite grating image. The composite grating image is programmed into the optical engine memory using optical engine programming software. The projection interval and projection mode of the optical engine are also set. Step 3: After the steel plate stops under the control of the metal switch, the moving gantry and vision system move to the first area of ​​the steel plate. The optomechanical system projects the image according to the burned composite grating image and projection settings. The camera is an industrial color camera, directly connected to the external trigger of the optomechanical system. After the optomechanical system projects the image, it triggers the camera to acquire the image. The composite grating image captured by the camera is transmitted to the computing server for phase calculation. The captured composite grating image is then processed through Butterworth bandpass and lowpass filters to recover the single grating image of each modulation channel. Step 4: Obtain the solved phase of a single grating image, and expand the solved phase into the absolute phase through spatial phase expansion; Step 5: Measure the irregular crack based on the calculated absolute phase; Step 6: Based on the discontinuities corresponding to the irregular crack locations in the calculated phase, establish the mapping relationship between the discontinuous phase and the three-dimensional point cloud of the crack, and calculate the three-dimensional coordinates of the crack based on the calculated discontinuous phase locations and the mapping relationship. Step 7: Based on the calculated 3D point cloud coordinates of the irregular crack, solve for the spatial location, length, and area of ​​the irregular crack. The spatial location of the irregular crack is represented by the geometric center and the direction vector. The geometric center represents the coordinate position of the crack, and the direction vector represents the spatial direction of the irregular crack.

[0015] Compared with the prior art, the beneficial effects of the present invention are as follows: This invention proposes a visual recognition and surface feature measurement method for irregular cracks in medium-thick plates. This method can identify irregular cracks in complete medium-thick plates and solve for the spatial location, length and area of ​​irregular cracks. It can provide information for realizing high-precision automated measurement of irregular cracks and for the next step of automatic crack repair process. This invention constructs a visual recognition and surface feature measurement system for irregular cracks in medium-thick plates. This system can identify irregular cracks on the surface of complete medium-thick plates and measure their spatial location, length, and area surface features. The information measured by this system will directly provide the surveyor with accurate crack surface information and feed the information back to the crack repair equipment, which will help the crack repair equipment to perform efficient and automatic repair. Attached Figure Description

[0016] To more clearly illustrate the technical solution of the present invention, the drawings used in the embodiments are briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0017] Figure 1 This is a flowchart illustrating the method for identifying irregular cracks in thick plates and measuring their spatial location, length, and area in an embodiment of the present invention. Figure 2 This is a structural diagram of the system for identifying irregular cracks in thick plates and measuring their spatial location, length, and area, as described in an embodiment of the present invention. Detailed Implementation

[0018] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0019] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0020] Example 1 The example uses the identification of irregular cracks in a medium-thick plate and the measurement of their spatial location, length, and area as an example. Figure 2 As shown, the present invention provides a visual recognition and surface feature measurement system for irregular cracks in medium-thick plates, comprising: The system comprises a vision system consisting of an optomechanical unit and a camera, a mobile gantry for mounting the vision system, and a system for calculating irregular cracks in a medium-thick plate. After the steel plate stops, it is divided into multiple regions based on its length and width dimensions and the minimum projection area of ​​the optomechanical unit in the vision system. The mobile gantry moves to the head of the plate, and the measurement system begins measuring from the first region, then moves region by region until the measurement is complete. The optomechanical unit in the measurement system projects a modulated and superimposed composite grating image onto the surface of the plate in one go. An externally triggered camera acquires the composite grating image modulated by the plate surface. By segmenting the spatial discontinuities in the phase map calculated from the captured composite grating image, the discontinuous coordinates of the irregular cracks in the absolute phase map are obtained. These discontinuous coordinates are then mapped to point cloud computing to calculate the length and area of ​​the irregular cracks.

[0021] Optionally, the vision system consisting of an optical engine and a camera includes the following functions: The optomechanical system capable of programming self-encoded projection images is set to external trigger projection mode. The self-encoded projection image is a computer-generated composite grating image. The composite grating image is programmed into the optomechanical memory using optomechanical programming software, and the projection interval and projection mode of the optomechanical system are set. An industrial color camera is used, directly connected to the external trigger of the optomechanical system. After the optomechanical system projects the image, it triggers the camera to acquire the image and send it to the medium-thick plate irregular crack calculation system.

[0022] Optionally, the method for generating the projected composite grating image is as follows: For a projected composite grating image, each individual grating image is modulated with a carrier pattern and superimposed together. The cumulative effect of each modulation pattern in the projected composite grating image is represented as: ; In the formula, This is a projected grating stripe image; The carrier frequency for projecting the grating stripe image; The horizontal coordinates of the projected grating stripe image; The phase is a three-step phase shift; and These are the background intensity and the reflected intensity, respectively. The number of projected grating stripe images.

[0023] Optionally, the functions of the mobile gantry crane for installing the vision system include: A metal switch is installed below the roller conveyor. When the steel plate is transported to the bottom of the roller conveyor, the metal switch is triggered to control the roller conveyor to stop moving. The steel plate is divided into multiple regions according to the length and width of the steel plate and the projection area of ​​the optical engine in the vision system. Then, the gantry is controlled to move to the head position of the steel plate. Taking the upper left corner of the steel plate as the projection starting point, the measurement is carried out region by region in order from left to right until the measurement is completed.

[0024] Optionally, the number of segmented regions in the medium-thick plate is .

[0025] The number of times the vision system moves along the length of the medium-thick plate is: ; The number of movements of the vision system in the width direction of the medium-thick plate is: ; in, This indicates the floor function, meaning rounding down to the nearest integer. and The smallest integer; the length and width of the medium-thick plate are respectively and The length of the minimum projection area of ​​the optical engine is and .

[0026] Optionally, the functions of the irregular crack calculation system for medium-thick plates include: The images captured by the camera are uploaded to the server. The single composite grating captured by the camera and modulated by the steel plate surface is phase-solved on the server, and discontinuous positions are segmented in the phase map. The discontinuous positions are mapped to the three-dimensional point cloud to obtain the area of ​​irregular cracks, and then visualized on the host computer.

[0027] Optionally, phase resolution of the composite grating image captured by the camera includes: The decoding process of a camera capturing a composite grating image is similar to the demodulation process in a wireless communication system. The captured image is passed through bandpass and lowpass filters to recover the pattern of each modulation channel. Each image after demodulation... This refers to a single raster image used to calculate the wrap phase, which can be represented as: ; In the formula, The average gray level of the raster image; This is the modulated grayscale value of the pixel; The phase to be wrapped; This represents the current phase shift; These are pixel coordinates.

[0028] The phase that can be solved from a single raster image is: ; in, For the demodulated first n Zhang raster image; For the first n The initial phase of a grating image.

[0029] Optionally, the solved phase is expanded into an absolute phase through spatial phase expansion, and the process of obtaining the discontinuities in the absolute phase includes: Absolute phase of the matching point The solution is: ; Among them, integer multiples That's the number of packages.

[0030] The least squares direction of the absolute phase diagram unfolded above The estimate is as follows: ; If discontinuities exist in the phase map, then two unequal eigenvalues ​​in the phase map are used to represent two different directional regions. and These are the eigenvectors corresponding to the two eigenvalues. A bilateral filter is applied to the least-squares pattern, and edges / textures in the output of the bilateral filter are extracted / removed. The output image is explicitly labeled into two regions: continuous regions with a value of 0 and discontinuous regions with a value of 1. Weight masks are obtained by relabeling these two regions; a weight mask of 0 corresponds to a phase-continuous region, and a weight mask of 1 corresponds to a phase-discontinuous region.

[0031] Optionally, based on the discontinuities corresponding to the irregular crack locations in the calculated phase, establishing the mapping relationship between the discontinuous phase and the three-dimensional point cloud of the crack includes: ; in, These are the three-dimensional point cloud coordinates of the crack; For mapping functions; The absolute phase of the matching point; For the rows of patterns projected by the projector; The frequency of the stripes in the projected pattern.

[0032] Optionally, based on the solved 3D point cloud information of the irregular crack, the surface features of the irregular crack are calculated as follows: The spatial location of an irregular crack is represented by its geometric center and direction vector. The geometric center represents the spatial location of the crack, and the direction vector represents the direction of the irregular crack. for: ; in, These represent the first and second irregular cracks. i Three-dimensional coordinate information of each point, Indicates the total number of irregular cracks n One point.

[0033] For the first irregular crack i There are points, and the direction vector of each point is: ; Optionally, based on the three-dimensional point cloud information of the irregular crack obtained above, the process for determining the length of the irregular crack is as follows: The 3D point cloud representation of irregular cracks is as follows , For the first irregular crack n The length of the irregular crack is calculated based on the three-dimensional coordinate information of each point and the cumulative distance of the path points. L for: ; Optionally, based on the obtained 3D point cloud information of the irregular crack, the process for calculating the area of ​​the irregular crack is as follows: Calculating the area enclosed by the 3D point cloud of an irregular crack requires reducing the 3D point cloud to a 2D plane using PCA projection. The PCA projection process includes data centralization, covariance matrix calculation, and PCA decomposition, as detailed below: ; in, For the centralized point cloud, The number of point clouds.

[0034] covariance matrix for: ; in, for The transpose of .

[0035] For covariance matrix The eigenvalue decomposition process is as follows: ; in, It is the characteristic matrix.

[0036] Select the first two principal components Using a 2D planar coordinate system as the basis, projecting 3D points onto this plane is represented as: ; For the projected 2D point Calculate the area enclosed by the points on the outer contour of the irregular crack using the following formula. ,Right now: .

[0037] This invention uses grating image processing to accurately locate discontinuous cracks in medium-thick plates and measure their spatial location, length, and area. The system does not require processing areas outside the cracks, ensuring rapid and accurate measurement of irregular cracks.

[0038] Example 2 like Figure 1 As shown, this invention discloses a method for visual recognition and surface feature measurement of irregular cracks in medium-thick plates, including the identification of irregular cracks in medium-thick plates, and the measurement of the spatial location, length, and area of ​​the irregular cracks. The identification of irregular cracks in medium-thick plates includes the following steps: Step 1: Generate the composite raster image shown below according to the following formula, which can be represented as: ; In the formula, This is a projected grating stripe image; The carrier frequency for projecting the grating stripe image; The horizontal coordinates of the projected grating stripe image; The phase is a three-step phase shift; and These are the background intensity and the reflected intensity, respectively. The number of projected grating stripe images.

[0039] Step 2: Set the optical engine with the ability to program self-encoded projection images to external trigger projection mode. The self-encoded projection image is a computer-generated composite grating image. Program the composite grating image into the optical engine memory using the optical engine programming software, and set the projection interval and projection mode of the optical engine.

[0040] Step 3: After the steel plate stops under the control of the metal switch, the moving gantry and vision system move to the first area of ​​the steel plate. The optomechanical system projects the image according to the burned composite grating image and projection settings. An industrial color camera is used, directly connected to the external trigger of the optomechanical system. After the optomechanical system projects the image, it triggers the camera to acquire the image. The composite grating image captured by the camera is transmitted to the computing server for phase calculation. The captured composite grating image is passed through Butterworth bandpass and lowpass filters to recover the single grating image of each modulation channel, which can be represented as: ; In the formula, The average gray level of the raster image; This is the modulated grayscale value of the pixel; The phase to be wrapped; This represents the current phase shift; These are pixel coordinates.

[0041] Step 4: The phase of a single raster image is calculated using the following formula: ; in, For the demodulated first n Zhang raster image; For the first n The initial phase of a grating image.

[0042] The solved phase is expanded into absolute phase using the spatial phase expansion of the following formula, and the absolute phase of the matching point is obtained. The solution is: ; Among them, integer multiples That's the number of packages.

[0043] Step 5: Measure the irregular crack based on the calculated absolute phase, specifically: The least-squares direction of the absolute phase diagram is calculated using the following formula. .

[0044] ; If discontinuities exist in the phase map, then two unequal eigenvalues ​​in the phase map are used to represent two different directional regions. and These are the eigenvectors corresponding to the two eigenvalues. A bilateral filter is applied to the least-squares pattern, and edges / textures in the output of the bilateral filter are extracted / removed. The output image is explicitly labeled into two regions: continuous regions with a value of 0 and discontinuous regions with a value of 1. Weight masks are obtained by relabeling these two regions; a weight mask of 0 corresponds to a phase-continuous region, and a weight mask of 1 corresponds to a phase-discontinuous region.

[0045] Step 6: Calculate the three-dimensional coordinates of the crack based on the calculated phase discontinuity locations and the following mapping relationship: ; in, These are the three-dimensional point cloud coordinates of the crack; For mapping functions; The absolute phase of the matching point; For the rows of patterns projected by the projector; The frequency of the stripes in the projected pattern.

[0046] Step 7: Based on the 3D point cloud coordinates of the irregular crack calculated in the previous steps and the following formulas, solve for the spatial location, length, and area of ​​the irregular crack. The spatial location of the irregular crack is represented by its geometric center and direction vector. The geometric center represents the coordinate position of the crack, and the direction vector represents the spatial direction of the irregular crack. for: ; in, These represent the first and second irregular cracks. i Three-dimensional coordinate information of each point, Indicates the total number of irregular cracks n One point; For the first irregular crack i There are points, and the direction vector of each point is: ; Based on the obtained 3D point cloud information of the irregular crack, the process for determining the length of the irregular crack is as follows: The 3D point cloud representation of irregular cracks is as follows , For the first irregular crack n The length of the irregular crack is calculated based on the three-dimensional coordinate information of each point and the cumulative distance of the path points. L for: ; Calculating the area enclosed by the 3D point cloud of an irregular crack requires reducing the 3D point cloud to a 2D plane using PCA projection. The PCA projection process includes data centralization, covariance matrix calculation, and PCA decomposition, as detailed below: ; in, For the centralized point cloud, The number of point clouds.

[0047] covariance matrix for: ; in, for The transpose of .

[0048] For covariance matrix The eigenvalue decomposition process is as follows: ; in, It is the characteristic matrix.

[0049] Select the first two principal components Using a 2D planar coordinate system as the basis, projecting 3D points onto this plane is represented as: ; For the projected 2D point Calculate the area enclosed by the points on the outer contour of the irregular crack using the following formula. ,Right now: .

[0050] At this point, the identification of irregular cracks in a single area and the measurement of their spatial location, length, and area are completed. The measurement system, in coordination with the gantry, moves to the next measurement area and repeats the above steps to begin measurement and continue until the measurement of the entire medium-thick plate is completed.

[0051] Using the above method, a new approach for visual recognition and surface feature measurement of irregular cracks in medium-thick plates was established. This method can overcome the shortcomings of irregular crack recognition and the difficulty in measuring the spatial location, length, and area of ​​cracks in medium-thick plates. It enables rapid identification and surface feature measurement of cracks in medium-thick plates and can be applied to the identification and measurement of surface crack features in metal surfaces such as steel plates and magnesium-aluminum alloy plates in the metallurgical industry. It has significant theoretical importance and substantial practical application value.

[0052] This embodiment takes the identification and spatial location, length, and area measurement of irregular cracks in medium-thick plates as an example. It includes a production line roller conveyor, a moving gantry, a vision system consisting of an optical engine and a camera, a server, and necessary power cables. The optical engine is a DLP4500, and the server uses a GTX4080 graphics card. After the steel plate stops, the moving gantry moves area by area to identify irregular cracks on the entire steel plate surface, and the irregular crack shape information is calculated based on the obtained three-dimensional point cloud information of the irregular cracks. This embodiment is a preferred implementation of this application, but the scope of protection of this application is by no means limited to this. It includes the types of medium-thick metal plates, the method for identifying irregular cracks, the method for calculating the spatial location, length, and area of ​​irregular cracks, the hardware configuration, and any easily conceivable modifications, all of which are covered within the scope of protection of this application.

[0053] The embodiments described above are merely preferred embodiments of the present invention and are not intended to limit the scope of the present invention. Various modifications and improvements made to the technical solutions of the present invention by those skilled in the art without departing from the spirit of the present invention should fall within the protection scope defined by the claims of the present invention.

Claims

1. A visual recognition and surface feature measurement system for irregular cracks in medium-thick plates, characterized in that, The system includes: a vision system consisting of an optical engine and a camera, a mobile gantry frame for mounting the vision system, and a calculation system for irregular cracks in medium-thick plates; After the steel plate stops, the medium-thick plate is divided into multiple areas according to the plate size and the minimum coverage area of ​​the vision system. The moving gantry is moved to the head position of the medium-thick plate, and the measurement system will start measuring from the first area of ​​the medium-thick plate and then move area by area until the measurement is completed. The optomechanical system in the measurement system projects the modulated and superimposed composite grating image onto the surface of the medium-thick plate in one go. The composite image modulated by the surface of the medium-thick plate is acquired by an external trigger camera. By segmenting the spatial discontinuities in the phase map obtained from the captured composite grating image, the discontinuity coordinates of the irregular crack in the absolute phase map of the medium-thick plate are obtained. The discontinuity is mapped to a point cloud to calculate the length and area of ​​the irregular crack in the medium-thick plate.

2. The system according to claim 1, characterized in that, A vision system consisting of an optical engine and a camera includes: Set the optical engine with a programmable self-encoded projection image to external trigger projection mode. The self-encoded projection image is a computer-generated projection composite grating image. The projection composite grating image is programmed into the optical engine memory through the optical engine programming software, and the projection interval time and projection mode of the optical engine are set. The camera is an industrial color camera, which is directly connected to the external trigger of the optomechanic. After the optomechanic projects the composite grating image, it will trigger the camera to acquire the image and send it to the calculation system for irregular cracks in medium and thick plates.

3. The system according to claim 1, characterized in that, The process of generating a composite raster image is as follows: For a composite grating image, each individual grating image is modulated with a carrier pattern and superimposed together. The composite grating image accumulates each modulation pattern, which is represented as: ； In the formula, This is a projected grating stripe image; The carrier frequency for projecting the grating stripe image; The horizontal coordinates of the projected grating stripe image; For a three-step phase shift, ; and These are the background intensity and the reflected intensity, respectively. The number of projected grating stripe images.

4. The system according to claim 1, characterized in that, The mobile gantry crane used for installing vision systems includes: A metal switch is installed below the roller conveyor. When the steel plate is transported to the bottom of the roller conveyor, the metal switch is triggered to control the roller conveyor to stop moving. The steel plate is divided into regions according to the steel plate size information. The size of the divided region is the projection area of ​​the optical engine. Then, the gantry is controlled to move to the head position of the steel plate. Taking the upper left corner of the steel plate as the projection starting point, the measurement is carried out region by region in order from left to right until the measurement is completed.

5. The system according to claim 1, characterized in that, The calculation system for irregular cracks in medium-thick plates includes: The images captured by the camera are uploaded to the server. The single composite grating captured by the camera and modulated by the steel plate surface is phase-solved on the server, and discontinuous positions are segmented in the phase map. The discontinuous positions are mapped to the three-dimensional point cloud to obtain the area of ​​irregular cracks, and then visualized on the host computer.

6. The system according to claim 5, characterized in that, The phase resolution process for composite grating images captured by a camera includes: The captured image is processed through bandpass and lowpass filters to recover the pattern of each modulation channel. The demodulated pattern is then used to calculate the wrapper phase, represented as: ； In the formula, The average gray level of the raster image; Modulated grayscale for pixels; The phase to be wrapped; This represents the current phase shift; These are pixel coordinates; The phase that can be solved using the obtained single raster image is: ； in, This is the nth raster image after demodulation; Let be the initial phase of the nth raster image.

7. The system according to claim 6, characterized in that, The process of expanding the solved phase into an absolute phase through spatial phase expansion, and then finding the discontinuous positions in the absolute phase, includes: Absolute phase of the matching point The solution is: ； Among them, integer multiples It is the number of packages; Least square direction of the unfolded absolute phase diagram The estimate is: ； If discontinuities exist in the phase map, then two unequal eigenvalues ​​in the phase map are used to represent two different directional regions. and These are the eigenvectors corresponding to the two eigenvalues; bilateral filtering is performed on the least squares pattern, and the edges / textures of the bilateral filter output are extracted / removed respectively. The output image is clearly marked as two regions: a continuous region with a value of 0 and a discontinuous region with a value of 1; by re-marking the two regions, a weight mask is obtained, that is, a weight mask of 0 corresponds to a phase continuous region, and a weight mask of 1 corresponds to a phase discontinuous region.

8. The system according to claim 7, characterized in that, Based on the discontinuities corresponding to the irregular crack locations in the calculated phase, the mapping relationship between the discontinuous phase and the three-dimensional point cloud of the crack is established, including: ； in, These are the three-dimensional point cloud coordinates of the crack; For mapping functions; The absolute phase of the matching point; For the rows of patterns projected by the projector; The frequency of the stripes in the projected pattern.

9. The system according to claim 8, characterized in that, Based on the obtained 3D point cloud information of the irregular crack, the calculation process of the surface features of the irregular crack is as follows: The spatial location of an irregular crack is represented by its geometric center and direction vector. The geometric center represents the spatial location of the crack, and the direction vector represents the direction of the irregular crack. for: ； in, These represent the three-dimensional coordinates of the i-th point of the irregular crack. This indicates that the irregular crack has a total of n points; Let i be the i-th point of the irregular crack, and let its direction vector be: ； Based on the obtained 3D point cloud information of the irregular crack, the process for determining the length of the irregular crack is as follows: The 3D point cloud representation of irregular cracks is as follows , Given the three-dimensional coordinates of the nth point of the irregular crack, the length L of the irregular crack is calculated based on the cumulative distance of the path points: ； Based on the obtained 3D point cloud information of the irregular crack, the process for calculating the area of ​​the irregular crack is as follows: Calculating the area enclosed by the 3D point cloud of an irregular crack requires reducing the 3D point cloud to a 2D plane using PCA projection. The PCA projection process includes data centralization, covariance matrix calculation, and PCA decomposition, as detailed below: ； in, For the centralized point cloud, The number of point clouds; covariance matrix for: ； in, for The transpose of the matrix; For covariance matrix The eigenvalue decomposition process is as follows: ； in, It is the characteristic matrix; Select the first two principal components Using a 2D planar coordinate system as the basis, projecting 3D points onto this plane is represented as: ； For the projected 2D point Calculate the area enclosed by the points on the outer contour of the irregular crack using the following formula. ,Right now: 。 10. A method for visual recognition and surface feature measurement of irregular cracks in medium-thick plates, said method being implemented using the system described in any one of claims 1-9, characterized in that... The method includes: Step 1: Obtain the composite grating image; Step 2: Set the optical engine with programmable self-encoded projection images to external trigger projection mode. The self-encoded projection image is a computer-generated projection composite grating image. The projection composite grating image is programmed into the optical engine memory using the optical engine programming software, and the projection interval time and projection mode of the optical engine are set. Step 3: After the steel plate stops under the control of the metal switch, the moving gantry and vision system move to the first area of ​​the steel plate. The optomechanical system projects the image according to the burned composite grating image and projection settings. The camera is an industrial color camera, directly connected to the external trigger of the optomechanical system. After the optomechanical system projects the image, it triggers the camera to acquire the image. The composite grating image captured by the camera is transmitted to the computing server for phase calculation. The captured composite grating image is then processed through Butterworth bandpass and lowpass filters to recover the single grating image of each modulation channel. Step 4: Obtain the solved phase of a single grating image, and expand the solved phase into the absolute phase through spatial phase expansion; Step 5: Measure the irregular crack based on the calculated absolute phase; Step 6: Based on the discontinuities corresponding to the irregular crack locations in the calculated phase, establish the mapping relationship between the discontinuous phase and the three-dimensional point cloud of the crack, and calculate the three-dimensional coordinates of the crack based on the calculated discontinuous phase locations and the mapping relationship. Step 7: Based on the calculated 3D point cloud coordinates of the irregular crack, solve for the spatial location, length, and area of ​​the irregular crack. The spatial location of the irregular crack is represented by the geometric center and the direction vector. The geometric center represents the coordinate position of the crack, and the direction vector represents the spatial direction of the irregular crack.

Images