Crack detection method and device for ribbed plate type part, electronic equipment and storage medium

Abstract

The invention provides a crack detection method and device for a ribbed plate type part, electronic equipment and a storage medium, and relates to the technical field of image processing. Acquiring a cost matrix of gradient amplitudes of pixel points in the laser scanning thermal image; constructing a path source matrix of the pixel points based on the cost matrix; based on the path source matrix, obtaining a feature retention mask of the pixel points to obtain a feature map of the laser scanning thermal image; and based on the feature map, analyzing the crack of the ribbed plate type part. According to the method, a dynamic planning feature path tracking algorithm is constructed by acquiring the cost matrix, the path source matrix and the feature retention mask of the pixel points, so that the separation of the crack and the interference signal of the rib plate area of the rib plate type part is successfully realized, and the accuracy of detecting the crack of the rib plate type part is improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of image processing, and in particular to a crack detection method and device for a ribbed plate type piece, an electronic device and a storage medium. BACKGROUND

[0002] At present, for high-reflective metal detection, most of them are to process the surface of the face material (for example, spray black paint, etc.) to improve the absorption rate for detection. In the existing high-reflective surface detection technology, the traditional morphological processing method suppresses interference through iterative operation of a structure element, but there are problems such as feature edge blurring and missing detection of small defects. Although the algorithm based on threshold segmentation can extract main features, it is sensitive to gradient intensity changes and is easily disturbed by local strong reflection, leading to feature breakage. In addition, the deep learning-based scheme relies on a large amount of labeled data, and the model generalization ability is weak, which is prone to false alarms in complex reflection scenarios, and has limitations such as high computing power requirement and high deployment cost. These methods are difficult to balance between feature integrity preservation, processing efficiency and anti-interference performance, which restricts the accuracy of high-reflective surface detection of ribbed plate configurations.

[0003] As a completely non-contact detection method, laser thermal imaging technology has been widely used in the field of non-destructive testing in recent years. Laser thermal imaging technology has unique advantages in the field of metal micro-crack detection. However, in traditional laser thermal imaging technology, the ribbed plate position of the ribbed plate type piece is easily determined as a crack feature, resulting in inaccurate crack detection of the ribbed plate type piece. SUMMARY

[0004] The present application provides a crack detection method and device for a ribbed plate type piece, an electronic device and a storage medium, to solve the defect that the crack detection of the ribbed plate type piece is not accurate in the prior art, and to improve the accuracy of crack detection of the ribbed plate type piece.

[0005] The present application provides a crack detection method for a ribbed plate type piece, comprising: acquiring a laser scanning thermal imaging image of the ribbed plate type piece; acquiring a cost matrix of gradient amplitudes of pixel points in the laser scanning thermal imaging image; constructing a path source matrix of the pixel points based on the cost matrix; based on the path source matrix, acquiring a feature retention mask of the pixel points to obtain a feature map of the laser scanning thermal imaging image; and based on the feature map, analyzing the crack of the ribbed plate type piece.

[0006] According to the crack detection method for the ribbed plate type piece provided by the present application, the path source matrix of the pixel points is constructed based on the cost matrix, comprising: in the cost matrix, the cumulative gradient amplitude from the starting pixel point of the laser scanning thermal imaging image to any pixel point is acquired, and based on the cumulative gradient amplitude of all pixel points, a cumulative cost matrix is obtained; in the cumulative cost matrix, the row coordinates of the adjacent minimum cumulative gradient amplitude of all pixel points in the cumulative cost matrix are determined based on a preset search range to obtain the path source matrix.

[0007] According to the crack detection method for ribbed plate parts provided by the present invention, the calculation formula of the cumulative gradient amplitude of the pixel point is as follows: ; in, Pixel The cumulative gradient amplitude, is the gradient amplitude of the pixel point, is the row coordinate of the pixel point, is the column coordinate of the pixel point, is the cumulative cost matrix The minimum cumulative gradient magnitude of the column within the preset search range, The row coordinates of the pixel The preset search range is The row coordinate value within the preset search range.

[0008] According to the crack detection method for ribbed plate parts provided by the present invention, the calculation formula for the row coordinates of the adjacent minimum cumulative gradient amplitudes of pixel points is as follows: ; in, is the row coordinate of the pixel point, The column coordinates of the pixel point are, is the row coordinate of the adjacent minimum cumulative gradient amplitude of the pixel point, The row coordinates of the pixel The preset search range is is the cumulative cost matrix The row coordinate of the minimum cumulative gradient magnitude within the preset search range of the column, The row coordinate value within the preset search range.

[0009] According to the crack detection method for a ribbed plate member provided by the present invention, a feature-preserving mask for a pixel point is obtained based on a path source matrix, including: obtaining an optimal row number for each column of pixels in a laser scanning thermal imaging image based on the path source matrix and a cumulative cost matrix; obtaining a feature-preserving mask for the pixel point based on the row coordinates of the pixel point and the optimal row number of a column of pixels matching the pixel point; the calculation formula for the feature-preserving mask for the pixel point is as follows: ; in, Pixel The feature preservation mask of For pixels Matching The optimal row number of the column pixel, is the row difference threshold.

[0010] The optimal row number of each column of pixels in the crack detection method of the ribbed plate type piece is calculated according to the following formula: ; Wherein, The optimal row number of the last column of pixels in the laser scanning thermal imaging image is, The maximum value of the row coordinates of the pixels is, The maximum value of the column coordinates of the pixels is, The optimal row number of the first column of pixels is, The optimal row number of the first column of pixels is, The row coordinate of the minimum cumulative gradient amplitude in the last column of the cumulative cost matrix is. The crack detection method of the ribbed plate type piece provided by the application obtains a laser scanning thermal imaging image of the ribbed plate type piece, including: based on a laser scanning thermal imaging system, scanning a uniformly rotating ribbed plate type piece to obtain multiple original local laser scanning thermal imaging images; based on the multiple original local laser scanning thermal imaging images, reconstructing an original global laser scanning thermal imaging image of the ribbed plate type piece; performing anisotropic diffusion filtering on the original global laser scanning thermal imaging image to obtain the laser scanning thermal imaging image.

[0011] The application also provides a crack detection device for a ribbed plate type piece, including: an acquisition module for acquiring a laser scanning thermal imaging image of the ribbed plate type piece; a first processing module for acquiring a cost matrix of gradient amplitudes of pixels in the laser scanning thermal imaging image; a second processing module for constructing a path source matrix of the pixels based on the cost matrix; a third processing module for acquiring a feature retention mask of the pixels based on the path source matrix, so as to acquire a feature map of the laser scanning thermal imaging image; and an analysis module for analyzing cracks of the ribbed plate type piece based on the feature map.

[0012] The application also provides an electronic device including a memory, a processor, and a computer program stored in the memory and executable on the processor, and the processor implements any of the crack detection methods of the ribbed plate type piece described above when executing the computer program.

[0013] The application also provides a non-transitory computer readable storage medium having a computer program stored thereon, and the computer program is executed by a processor to implement any of the crack detection methods of the ribbed plate type piece described above.

[0014]

[0015] ​The application provides a crack detection method and device for a ribbed plate type piece, an electronic device and a storage medium. BRIEF DESCRIPTION OF DRAWINGS

[0016] In order to more clearly illustrate the technical solutions in the application or the prior art, the drawings needed in the embodiments or the prior art description will be briefly introduced. Obviously, the drawings in the following description are some embodiments of the application, and other drawings can be obtained by those skilled in the art without creative labor.

[0017] Figure 1 FIG. 1 is a flowchart of a crack detection method for a ribbed plate type piece provided by the application.

[0018] Figure 2 FIG. 2 is a structural diagram of a ribbed plate type piece provided by the application.

[0019] Figure 3 FIG. 3 is a scanning route diagram of a ribbed plate type piece provided by the application.

[0020] Figure 4 FIG. 4 is an original local laser scanning thermal imaging diagram provided by the application.

[0021] Figure 5 FIG. 5 is a time sequence temperature peak coordinate diagram provided by the application.

[0022] Figure 6 FIG. 6 is a flowchart of synthesizing an original global laser scanning thermal imaging diagram from an original local laser scanning thermal imaging diagram provided by the application.

[0023] Figure 7 (a) is a thermal response distribution diagram of an original local laser scanning thermal imaging diagram provided by the application.

[0024] Figure 7 (b) is a thermal response distribution diagram of position one provided by the application.

[0025] Figure 7 (c) is a thermal response distribution diagram of position two provided by the application.

[0026] Figure 7 (d) is the temperature gradient map provided by the present application in position two.

[0027] Figure 8 (a) is the gradient map of the original local laser scanning thermal imaging map provided by the present application.

[0028] Figure 8 (b) is the gradient map of the feature map provided by the present application.

[0029] Figure 9 (a) is the original global laser scanning thermal imaging map synthesized according to the original local laser scanning thermal imaging map provided by the present application.

[0030] Figure 9 (b) is the feature map obtained after the DP-FPT algorithm processing provided by the present application.

[0031] Figure 10 is the signal-to-noise ratio analysis result of the original global laser scanning thermal imaging map and the feature map obtained after the DP-FPT algorithm processing provided by the present application according to different processes.

[0032] Figure 11 is the structural schematic diagram of the crack detection device for the ribbed plate type part provided by the present application.

[0033] Figure 12 is the structural schematic diagram of the electronic device provided by the present application. DETAILED DESCRIPTION

[0034] In order to make the purpose, technical scheme and advantages of the present application clearer, the technical scheme in the present application will be described clearly and completely below in combination with the drawings in the present application. Obviously, the described embodiments are part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor fall within the scope of protection of the present application.

[0035] The crack detection method, device and electronic device for the ribbed plate type part of the present application will be described below in combination with Figures 1-12

[0036] Figure 1 is the flowchart of the crack detection method for the ribbed plate type part provided by the present application, as Figure 1 shown, the crack detection method for the ribbed plate type part includes steps S100 to S500, and each step is specifically as follows.

[0037] S100: Obtain the laser scanning thermal imaging map of the ribbed plate type part.

[0038] ​To realize the characterization of micro-cracks around the stringer panel of large curvature surface parts, the stringer panel type part of the present application includes artificial simulation samples. For example, the stringer panel type part is tightly attached by 9 blocks of aluminum alloy, and the micro-cracks are simulated by the gap between the laminated surfaces. The stringer panel type part is made of 6601 aluminum alloy, and the size is as shown in Figure 2 The laminated surface of each aluminum alloy block is polished, the misaligned joints between the aluminum alloy blocks are spliced, fixed by positioning pins, and tightly fitted by screws. The surface of the stringer panel type part of the present application is not pretreated, for example, by black paint spraying. The present application mainly studies the horizontal cracks near the horizontal stringer panel of the stringer panel type part.

[0039] A laser scanning thermal imaging system is established, which includes a continuous wave (CW) laser heat source, an infrared thermal imaging unit and a motion control platform. The working principle of the laser scanning thermal imaging system can be decomposed into three cooperative processes: first, a continuous wave laser with a wavelength of 808 nm and a maximum power of 100 W is used as a thermal excitation source, and by adjusting the combination of the laser excitation current and the fiber collimator, the dynamic regulation of the laser excitation power and the spot size is realized. Second, the rotation angle, displacement stroke and motion speed of the motion control platform are accurately controlled by the digital hardware controller (DHC) module, so that the stringer panel type part (measured part) fixed on the motion control platform realizes programmable scanning motion. Finally, a cooled mid-wave infrared thermal imager with a spatial resolution of 512×640 pixels and a maximum frame rate of 120 Hz is used to synchronously collect the thermal response signal. It is worth noting that the fiber collimator and the infrared thermal imager remain spatially stationary during the detection process, and only the motion platform drives the test piece to complete the relative displacement scanning.

[0040] The scanning route is as shown in Figure 3 Lines A, B and C are respectively the crack position, 1 mm below the crack and 2 mm below the crack. The present application adopts a double laser power excitation scheme (30 W, 45 W), configures the thermal imager to collect data with a left bias of 10° and a depression angle of 15°, and the acquisition frequency is 60 Hz. The motion control platform moves at a constant speed of 5.86 mm / s to complete the laser scanning thermal imaging data acquisition, and a plurality of original local laser scanning thermal imaging images are obtained, as shown in Figure 4 The original local laser scanning thermal imaging images are combined to obtain an original global laser scanning thermal imaging image, as shown in Figure 6

[0041] The time sequence temperature peak value coordinate graph of the original local laser scanning thermal imaging image is obtained, as shown in Figure 5 The thermal response distribution graph and the temperature gradient graph of the original local laser scanning thermal imaging image are obtained, as shown in Figure 7 ​(a), 7(b), 7(c), and 7(d).

[0042] The original local laser scanning thermography image usually has the following problems. (1) The mirror reflection artifact generated by the horizontal rib plate of the rib plate type part presents a similar pseudo-thermal response to the real crack, which is caused by the close distribution of the crack initiation zone and the horizontal rib plate (e.g., the distance between the crack and the horizontal rib plate is <5 mm) and the high reflection characteristics of the aluminum alloy substrate (reflectivity >85% at 808 nm wavelength). (2) The secondary thermal anomaly in the area to the right of the excitation point has a clear temperature gradient and a strong spatial correlation with the rib plate geometry, and a temperature anomaly cold zone also appears. The crack location and the rib plate area also have a gradient change feature. (3) The heat accumulation phenomenon is significantly enhanced in the chamfered part of the rib plate type part due to the local heat flow distortion caused by the edge heat accumulation effect, and obvious heat accumulation phenomenon appears, and the high-temperature high-brightness elongated straight line is easily mistaken for a crack. (4) The temperature peak coordinate graph is shown in Figure 5 , where X and Y represent the coordinate changes of the temperature peak points in the space position during the entire scanning process. As shown in Figure 4 , the thermal signal spatial drift phenomenon appears in the area adjacent to the laser excitation point. Due to the rib plate reflection and chamfer interference, the temperature peak coordinate on the rib plate surface presents irregular jumping.

[0043] As shown in Figure 4 , 7 (a), 7(b), 7(c), the experimental data shows that in the thermal response distribution graph, there are two abnormal temperature peak areas outside the crack area, which are the rib plate position and the chamfer position, and the thermal response peak temperature of the rib plate area is only 0.88% different from that of the crack area. The thermal response peak temperature of the chamfer position is 2.13% different from that of the crack area. After thermal image reconstruction, the rib plate area still presents a significant high-temperature response, and the relative error of the temperature peak value of the rib plate area and the crack area is less than 11.4%, and the gradient characteristics of the pseudo-defect in the rib plate area have a high similarity in spatial distribution form with the real crack, but the thermal response of the chamfer position is relatively reduced, and the influence on the actual detection process is small. It is worth noting that the temperature peak of the rib plate interference area in some detection positions even exceeds the crack area by 22.02%, resulting in a pseudo-defect feature gradient in the corresponding temperature gradient graph that is 109.8% higher than the real crack gradient. Such pseudo-defect features have a high degree of convergence in spatial distribution and amplitude characteristics with the real crack, making it easy to misjudge by traditional thermal imaging analysis methods, significantly increasing the difficulty of crack feature extraction. In the data processing process, the feature at the rib plate position is easily mistaken for a crack feature, which greatly increases the detection difficulty, making it extremely difficult to extract micro-cracks by traditional processing methods.

[0044] In order to eliminate interference and effectively extract crack features, a dynamic programming fixed-parameter tractable (DP-FPT) algorithm is designed to process the original global laser scanning thermal imaging image.

[0045] Firstly, the original global laser scanning thermal imaging image is filtered to obtain a laser scanning thermal imaging image of the ribbed plate type part.

[0046] S200: Obtain a cost matrix of gradient amplitudes of pixel points in the laser scanning thermal imaging image.

[0047] The calculation formula of the gradient amplitude of each pixel point in the cost matrix is as follows.

[0048] ; Wherein, is a matrix value corresponding to the pixel point in the cost matrix, is the gradient amplitude of the pixel point .

[0049] The gradient amplitude of each pixel point in the laser scanning thermal imaging image is taken as a matrix value to obtain the corresponding cost matrix.

[0050] S300: Construct a path source matrix of pixel points based on the cost matrix.

[0051] The application converts the maximum gradient path problem into a minimum cumulative cost problem, and the high gradient area corresponds to a smaller cost value, which guides the path tracking main feature. The cost matrix directly determines the direction of the path, and the cost gradient matrix of the application adopts a negative gradient design to ensure that the algorithm preferentially selects the high gradient area and effectively avoids the interference of the higher gradient at the rib plate.

[0052] According to the cost matrix, the cumulative cost matrix is calculated by dynamic programming recursion. According to the cumulative cost matrix, the path source matrix is constructed. The path source matrix is used to store the optimal decision history in the forward calculation process of dynamic programming, and provides chain mapping for reverse reconstruction of the path in the backtracking stage.

[0053] The application adopts a negative gradient cost matrix construction method to convert high gradient features into a minimum path problem, and effectively highlights the main feature line of the crack through reverse cumulative cost calculation.

[0054] S400: Based on the path source matrix, obtain a feature reservation mask of the pixel points to obtain a feature map of the laser scanning thermal imaging image.

[0055] According to the path source matrix inverse reconstruction complete path, the numerical information in the accumulated cost matrix is converted into continuous paths in the physical space, and then the feature reservation mask of the pixel points is generated based on the paths.

[0056] Finally, the feature reservation mask of all pixel points is median filtered to eliminate residual noise, and the feature map of the laser scanning thermal imaging image is obtained.

[0057] S500: Based on the feature map, analyze the crack of the ribbed plate type part.

[0058] Analyze the feature map to identify the crack of the ribbed plate type part. For example, obtain the gradient map of the feature map, as shown in Figure 8 (b). According to the peak value in the gradient map of the feature map, the crack of the ribbed plate type part is identified.

[0059] As shown in Figure 8 (a), in the gradient map of the original local laser scanning thermal imaging image, there are interferences of the rib plate and the chamfer. As shown in Figure 8 (b), in the gradient map of the feature map obtained after processing by the DP-FPT algorithm, the interferences of the rib plate and the chamfer are eliminated, and only the gradient peak value of the crack is reserved.

[0060] As shown in Figure 9 (a), in the original global laser scanning thermal imaging image synthesized by the original local laser scanning thermal imaging image, there are obvious interferences of the rib plate and the chamfer. As shown in Figure 9 (b), in the feature map obtained after processing by the DP-FPT algorithm, the interferences of the rib plate and the chamfer are eliminated.

[0061] Figure 10 is the signal-to-noise ratio analysis result of the original global laser scanning thermal imaging image obtained according to different processes and the feature map obtained after processing by the DP-FPT algorithm provided by the present application. As shown in Figure 10 It can be seen that the signal-to-noise ratio of the feature map is obviously higher than that of the original global laser scanning thermal imaging image.

[0062] The present application realizes effective identification of the main path and the reflection interference path of the crack through cost decision, and the crack feature extraction accuracy is more than 95% under the condition of 45W laser direct excitation at the crack position, the signal-to-noise ratio is improved by 27.2dB. The feasibility of laser infrared thermal imaging technology for crack detection in complex configurations and high reflection and low absorption materials is improved.

[0063] The crack detection method of the ribbed plate type piece provided by the embodiment of the present application comprises the following steps: obtaining a laser scanning thermal imaging image of the ribbed plate type piece; obtaining a cost matrix of gradient amplitudes of pixel points in the laser scanning thermal imaging image; constructing a path source matrix of the pixel points based on the cost matrix; obtaining a feature reservation mask of the pixel points based on the path source matrix, so as to obtain a feature map of the laser scanning thermal imaging image; and analyzing the crack of the ribbed plate type piece based on the feature map. The feature path tracking algorithm of dynamic programming is constructed by obtaining the cost matrix, the path source matrix and the feature reservation mask of the pixel points, the interference signal of the crack and the ribbed plate region of the ribbed plate type piece is successfully separated, and the accuracy of detecting the crack of the ribbed plate type piece is improved.

[0064] The purpose of the present application is to detect the crack at the ribbed plate of the uncoated high-curvature high-reflectivity ribbed plate type piece by laser infrared thermal imaging technology. Due to the high-reflectivity characteristics of the material and the complex reflection of the high curvature, the ribbed plate region appears a similar thermal response feature to the crack position, the thermal signal temperature amplitudes of the crack and the ribbed plate region are very small, the reflection interference phenomenon is serious, the feature at the crack is not a continuous high-amplitude signal, but presents a discontinuous high-low fluctuation, and the traditional threshold method is difficult to effectively extract the crack. In order to remove the interference of the ribbed plate region and effectively extract the crack signal, the feature path tracking algorithm of dynamic programming is successfully used to separate the crack and the interference signal.

[0065] Based on the above embodiment, the laser scanning thermal imaging image of the ribbed plate type piece is obtained, which comprises the following steps: Based on the laser scanning thermal imaging system, a plurality of original local laser scanning thermal imaging images are obtained by scanning the uniformly rotating ribbed plate type piece; Based on the plurality of original local laser scanning thermal imaging images, an original global laser scanning thermal imaging image of the ribbed plate type piece is reconstructed; The original global laser scanning thermal imaging image is subjected to anisotropic diffusion filtering to obtain the laser scanning thermal imaging image.

[0066] The ribbed plate type piece is placed on the motion control platform of the laser scanning thermal imaging system to realize the uniform rotation of the ribbed plate type piece. According to the laser scanning thermal imaging system constructed above, a plurality of original local laser scanning thermal imaging images are obtained by scanning the uniformly rotating ribbed plate type piece. The plurality of original local laser scanning thermal imaging images are reconstructed to obtain an original global laser scanning thermal imaging image.

[0067] Firstly, the original global laser scanning thermal imaging image is subjected to anisotropic diffusion filtering to obtain the laser scanning thermal imaging image. The formula of the anisotropic diffusion filtering is as follows.

[0068] ; wherein, a pixel point of a laser scanning thermal imaging image obtained after anisotropic diffusion filtering, a displacement of a row coordinate of the pixel point, a displacement of a vertical coordinate of the pixel point, a convolution window size, a control kernel function decay rate, a pixel point of an original global laser scanning thermal imaging image the pixel point after the displacement.

[0069] The anisotropic diffusion filtering of the application eliminates random high gradient points such as mirror reflection interference at a muscle plate by Gaussian kernel weighted average, ensures that the DP-FPT algorithm searches in a low noise environment, and avoids path deviation caused by noise.

[0070] Based on the above embodiment, a path source matrix of the pixel point is constructed based on the cost matrix, including the following steps: In the cost matrix, the cumulative gradient amplitude from a starting pixel point of the laser scanning thermal imaging image to any pixel point is obtained, and based on the cumulative gradient amplitudes of all the pixel points, a cumulative cost matrix is obtained. In the cumulative cost matrix, the row coordinates of the adjacent minimum cumulative gradient amplitudes of all the pixel points in the cumulative cost matrix are determined based on a preset search range, and a path source matrix is obtained.

[0071] Specifically, the calculation formula of the cumulative gradient amplitude of the pixel point is as follows: ; wherein, the cumulative gradient amplitude of the pixel point, the gradient amplitude of the pixel point, the row coordinate of the pixel point, the column coordinate of the pixel point, the minimum cumulative gradient amplitude of the pixel point in the preset search range, the row coordinate of the pixel point the preset search range, the value of the row coordinate in the preset search range. The cumulative gradient amplitudes of the pixel points in the second column to the last column are obtained by summing the current gradient amplitude of the pixel point and the minimum gradient amplitude of the previous column in the preset search range. Based on the cumulative gradient amplitudes of all the pixel points, a cumulative cost matrix is obtained.

[0072] Further, the cumulative gradient amplitude of the pixel point in the first column is equal to the current gradient amplitude of the pixel point in the first column.

[0073] Further, the cumulative gradient amplitude of the pixel point in the first column is equal to the current gradient amplitude of the pixel point in the first column.

[0074] ​The application obtains the accumulated gradient amplitude of the target pixel point according to the sum of the current gradient amplitude of the pixel point and the minimum gradient amplitude of the previous column in the preset search range, and realizes that the path from the starting point to the current position of each pixel point belongs to the global optimum.

[0075] Specifically, the calculation formula of the row coordinate of the adjacent minimum accumulated gradient amplitude of the pixel point is as follows: ; Among them, is the row coordinate of the pixel point, is the column coordinate of the pixel point, is the row coordinate of the adjacent minimum accumulated gradient amplitude of the pixel point, is the row coordinate of the pixel point is the preset search range, is the row coordinate of the minimum accumulated gradient amplitude in the preset search range of the first column of the accumulated cost matrix, is the value of the row coordinate in the preset search range.

[0076] For example, the accumulated cost matrix is as follows.

[0077] .

[0078] In order to specifically illustrate the embodiments of the application, the accumulated cost matrix in the example is the simplified accumulated cost matrix, and the number of rows and columns of the accumulated cost matrix in the actual data processing process is more.

[0079] For example, the preset search range is , and . The accumulated cost matrix For example, column 2. For the row coordinate of the adjacent minimum cumulative gradient amplitude of the first row of column 2, the minimum cumulative gradient amplitude can only be searched from the first row and the second row of column 1 (because the search range of the row coordinate is 2, 1 and-1, but there is no-1 row, so the minimum cumulative gradient amplitude can only be searched from the first row and the second row of column 1). The cumulative gradient amplitude of the first row of column 1 is 2.1, the cumulative gradient amplitude of the second row of column 1 is 1.8, and the minimum cumulative gradient amplitude is 1.8. The row coordinate of the minimum cumulative gradient amplitude 1.8 is 2, so p(1, 2) = 2. For the row coordinate of the adjacent minimum cumulative gradient amplitude of the second row of column 2, the minimum cumulative gradient amplitude is searched from the first row, the second row and the third row of column 1. The cumulative gradient amplitude of the first row of column 1 is 2.1, the cumulative gradient amplitude of the second row of column 1 is 1.8, and the cumulative gradient amplitude of the third row of column 1 is 2.5. The minimum cumulative gradient amplitude is 1.8. The row coordinate of the minimum cumulative gradient amplitude 1.8 is 2, so p(2, 2) = 2.

[0080] Further, in the path source matrix, the value of the first column is empty.

[0081] According to the above method, the row coordinates of the adjacent minimum cumulative gradient amplitudes of all pixel points are calculated , and a path source matrix is obtained.

[0082] .

[0083] The present application determines the row coordinates of the adjacent minimum cumulative gradient amplitudes of the pixel points by the row coordinates of the minimum cumulative gradient amplitudes in the preset search range of the first column of the cumulative cost matrix, and realizes the recording of the previous row coordinates of the optimal path reaching each pixel point.

[0084] The present application obtains a cumulative cost matrix according to a cost matrix, constructs a path source matrix according to the cumulative cost matrix, realizes a feature path tracking algorithm for dynamic programming of a laser scanning thermal imaging image, realizes effective identification of a main path and a reflection interference path through cost decision, and is beneficial to improving the accuracy of laser infrared thermal imaging technology in crack detection in complex configurations and high reflection and low absorption materials.

[0085] Based on the above embodiment, based on the path source matrix, a feature retention mask of a pixel point is obtained, including the following steps: Based on the path source matrix and the cumulative cost matrix, the optimal row number of each column of pixel points in the laser scanning thermal imaging image is obtained; Based on the row coordinate of the pixel point and the optimal row number of the column of pixel points matched with the pixel point, the feature retention mask of the pixel point is obtained; ​The calculation formula of the feature reservation mask of the pixel point is as follows: ; Wherein, is the feature reservation mask of the pixel point , is the optimal row number of the pixel point in the th column matched with the pixel point , and is a row difference threshold. Specifically, the calculation formula of the optimal row number of each column of pixel points is as follows:

[0086] ; Wherein, is the optimal row number of the last column of pixel points in the laser scanning thermal imaging image , is the maximum value of the row coordinates of the pixel points, is the maximum value of the column coordinates of the pixel points, is the optimal row number of the pixel points in the th column, and is the row coordinate of the minimum cumulative gradient amplitude in the last column of the cumulative cost matrix. For example, the cumulative cost matrix and the path source matrix

[0087] are as follows.

[0088] , .

[0089] .

[0090] .

[0091] That is, the value of the 4th row and the 5th column of the path source matrix is queried. For example, the value of the 4th row and the 5th column of the path source matrix is 4. According to the path source matrix and the cumulative cost matrix, the present application realizes accurate calculation of the optimal row number of the pixel point. The present application designs a column-directional strip mask generation algorithm, dynamically generates a binary template through path backtracking coordinates, and realizes accurate spatial separation of the feature band of the crack and the interference area.

[0092]

[0093]

[0094] ​​​​​​​The application realizes separation of the interference signal of the crack and the rib plate area of the rib plate type piece by generating the feature reservation mask of the pixel point, and improves the accuracy of detecting the crack of the rib plate type piece.

[0095] Further, the feature reservation mask of all pixel points is subjected to median filtering to eliminate residual noise, and the feature map of the laser scanning thermal imaging image is obtained. The median filtering can improve the smoothness of the feature map, and ensure that the crack feature is clean and visible.

[0096] ; wherein, is the pixel point of the feature map obtained after the DP-FPT algorithm processing, is the feature reservation mask of the pixel point .

[0097] The crack detection device for the rib plate type piece provided by the application will be described below. The crack detection device for the rib plate type piece described below can be referred to in correspondence with the crack detection method for the rib plate type piece described above.

[0098] As shown in Figure 11 , a crack detection device for a rib plate type piece includes an acquisition module 601, a first processing module 602, a second processing module 603, a third processing module 604, and an analysis module 605.

[0099] The acquisition module 601 is configured to acquire a laser scanning thermal imaging image of the rib plate type piece.

[0100] The first processing module 602 is configured to acquire a cost matrix of gradient amplitudes of pixel points in the laser scanning thermal imaging image.

[0101] The second processing module 603 is configured to construct a path source matrix of the pixel points based on the cost matrix.

[0102] The third processing module 604 is configured to acquire a feature reservation mask of the pixel points based on the path source matrix, so as to acquire a feature map of the laser scanning thermal imaging image.

[0103] The analysis module 605 is configured to analyze the crack of the rib plate type piece based on the feature map.

[0104] The crack detection device for ribbed plate parts provided in an embodiment of the present invention obtains a laser scanning thermal image of the ribbed plate part; obtains a cost matrix of the gradient amplitudes of pixels in the laser scanning thermal image; constructs a path source matrix for the pixels based on the cost matrix; obtains a feature-preserving mask for the pixels based on the path source matrix to obtain a feature map of the laser scanning thermal image; and analyzes cracks in the ribbed plate part based on the feature map. By obtaining the cost matrix, the path source matrix, and the feature-preserving mask for the pixels, the present invention constructs a dynamic programming feature path tracking algorithm, successfully separating cracks from interference signals in the ribbed plate region of the ribbed plate part, and improving the accuracy of crack detection in the ribbed plate part.

[0105] In one embodiment, the second processing module 603 is used to: obtain the cumulative gradient amplitude from the starting pixel point to any pixel point of the laser scanning thermal imaging image in the cost matrix, and obtain the cumulative cost matrix based on the cumulative gradient amplitudes of all pixel points; in the cumulative cost matrix, determine the row coordinates of the adjacent minimum cumulative gradient amplitudes of all pixel points in the cumulative cost matrix based on a preset search range to obtain the path source matrix.

[0106] In one embodiment, the calculation formula for the cumulative gradient amplitude of a pixel is as follows: ; in, Pixel The cumulative gradient amplitude, is the gradient amplitude of the pixel point, is the row coordinate of the pixel point, is the column coordinate of the pixel point, is the cumulative cost matrix The minimum cumulative gradient magnitude of the column within the preset search range, The row coordinates of the pixel The preset search range is The row coordinate value within the preset search range.

[0107] In one embodiment, the calculation formula for the row coordinates of the adjacent minimum cumulative gradient magnitudes of a pixel point is as follows: ; in, is the row coordinate of the pixel point, The column coordinates of the pixel point are, is the row coordinate of the adjacent minimum cumulative gradient amplitude of the pixel point, The row coordinates of the pixel The preset search range is is the cumulative cost matrix The row coordinate of the minimum cumulative gradient magnitude within the preset search range of the column, The row coordinate value within the preset search range.

[0108] In one embodiment, the third processing module 604 is configured to: obtain the optimal row number for each column of pixels in the laser scanning thermal image based on the path source matrix and the cumulative cost matrix; obtain a feature preservation mask for the pixel based on the row coordinates of the pixel and the optimal row number of a column of pixels that matches the pixel; the feature preservation mask for the pixel is calculated using the following formula: ; in, Pixel The feature preservation mask of For pixels Matching The optimal row number of the column pixel, is the row difference threshold.

[0109] In one embodiment, the optimal row number for each column of pixels is calculated as follows: ; in, The last column of the laser scanning thermal image The optimal row number of the pixel, is the maximum value of the row coordinate of the pixel point, is the maximum value of the column coordinates of the pixel point, For the The optimal row number of the column pixel, is the row coordinate of the minimum cumulative gradient magnitude in the last column of the cumulative cost matrix.

[0110] In one embodiment, the acquisition module 601 is used to: scan the uniformly rotating ribbed plate-shaped component based on a laser scanning thermal imaging system to obtain multiple original local laser scanning thermal images; reconstruct the original global laser scanning thermal image of the ribbed plate-shaped component based on the multiple original local laser scanning thermal images; and perform anisotropic diffusion filtering on the original global laser scanning thermal image to obtain a laser scanning thermal image.

[0111] Figure 12 An example of a physical structure diagram of an electronic device is shown below. Figure 12As shown, the electronic device can include a processor 710, a communications interface 720, a memory 730, and a communications bus 740, wherein the processor 710, the communications interface 720, and the memory 730 complete mutual communication through the communications bus 740. The processor 710 can invoke a logic instruction in the memory 730 to execute the crack detection method of the ribbed plate type piece, and the method includes: acquiring a laser scanning thermal imaging image of the ribbed plate type piece; acquiring a cost matrix of gradient amplitudes of pixel points in the laser scanning thermal imaging image; constructing a path source matrix of the pixel points based on the cost matrix; based on the path source matrix, acquiring a feature reservation mask of the pixel points to acquire a feature map of the laser scanning thermal imaging image; and based on the feature map, analyzing the crack of the ribbed plate type piece.

[0112] In addition, the logic instruction in the memory 730 described above can be implemented in the form of a software function unit and sold or used as an independent product, and can be stored in a computer readable storage medium. Based on such understanding, the technical solutions of the present application essentially or the part that contributes to the prior art or part of the technical solutions can be embodied in the form of a software product, and the computer software product is stored in a storage medium, including a plurality of instructions to make a computer device (which can be a personal computer, a server, or a network device, etc.) execute all or part of the steps of the method described in various embodiments of the present application. The foregoing storage medium includes: a U disk, a mobile hard disk, a read-only memory (ROM, Read-Only Memory), a random access memory (RAM, Random Access Memory), a magnetic disk or an optical disk, and various media that can store program codes.

[0113] In another aspect, the present application also provides a non-transitory computer readable storage medium having a computer program stored thereon, wherein the computer program is executed by a processor to implement the crack detection method of the ribbed plate type piece provided by the above-mentioned method, and the method includes: acquiring a laser scanning thermal imaging image of the ribbed plate type piece; acquiring a cost matrix of gradient amplitudes of pixel points in the laser scanning thermal imaging image; constructing a path source matrix of the pixel points based on the cost matrix; based on the path source matrix, acquiring a feature reservation mask of the pixel points to acquire a feature map of the laser scanning thermal imaging image; and based on the feature map, analyzing the crack of the ribbed plate type piece.

[0114] The device embodiments described above are merely illustrative, wherein the units described as separate components can or can not be physically separate, and the components displayed as units can or can not be physical units, i.e., can be located in one place, or can be distributed to multiple network units. Part or all of the modules can be selected to achieve the purposes of the embodiments according to actual needs. Those skilled in the art can understand and implement without creative labor.

[0115] Through the description of the above embodiments, those skilled in the art can clearly understand that the embodiments can be realized by means of software and the necessary general hardware platform, and of course can also be realized by hardware. Based on such understanding, the above technical solutions can be embodied in the form of a software product, which can be stored in a computer readable storage medium, such as a ROM / RAM, a magnetic disk, an optical disk, etc., and includes a number of instructions to make a computer device (which can be a personal computer, a server, or a network device, etc.) execute the methods described in each embodiment or some parts of the embodiments.

[0116] Finally, it should be noted that: the above embodiments are only used to illustrate the technical solutions of the present application, and not to limit them; although the present application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that: it can still modify the technical solutions recorded in the foregoing embodiments, or make equivalent replacement for part of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims

1. A crack detection method for a ribbed plate-shaped part, characterized in that: include: Obtain laser scanning thermal imaging images of ribbed plate parts; Obtaining a cost matrix of gradient amplitudes of pixel points in the laser scanning thermal imaging image; Constructing a path source matrix of the pixel points based on the cost matrix; Based on the path source matrix, a feature preservation mask of the pixel points is obtained to obtain a feature map of the laser scanning thermal imaging image; Based on the characteristic diagram, cracks in the ribbed plate-shaped component are analyzed.

2. The crack detection method for a ribbed plate-shaped part according to claim 1, characterized in that: The step of constructing the path source matrix of the pixel points based on the cost matrix includes: In the cost matrix, the cumulative gradient amplitude from the starting pixel point to any pixel point of the laser scanning thermal imaging image is obtained, and the cumulative cost matrix is ​​obtained based on the cumulative gradient amplitudes of all pixel points; In the cumulative cost matrix, row coordinates of adjacent minimum cumulative gradient magnitudes of all pixels in the cumulative cost matrix are determined based on a preset search range to obtain the path source matrix.

3. The crack detection method for a ribbed plate-shaped part according to claim 2, characterized in that: The calculation formula for the cumulative gradient amplitude of the pixel point is as follows: ； in, Pixel The cumulative gradient amplitude, is the gradient amplitude of the pixel point, is the row coordinate of the pixel point, is the column coordinate of the pixel point, is the cumulative cost matrix The minimum cumulative gradient magnitude of the column within the preset search range, The row coordinates of the pixel The preset search range is The row coordinate value within the preset search range.

4. The crack detection method for a ribbed plate-shaped part according to claim 2, characterized in that: The calculation formula for the row coordinates of the adjacent minimum cumulative gradient amplitude of the pixel point is as follows: ； in, is the row coordinate of the pixel point, The column coordinates of the pixel point are, is the row coordinate of the adjacent minimum cumulative gradient amplitude of the pixel point, The row coordinates of the pixel The preset search range is is the cumulative cost matrix The row coordinate of the minimum cumulative gradient magnitude within the preset search range of the column, The row coordinate value within the preset search range.

5. The crack detection method for a ribbed plate-shaped part according to claim 2, characterized in that: The step of obtaining a feature-preserving mask of a pixel point based on the path source matrix includes: Obtaining an optimal row number for each column of pixels in the laser scanning thermal imaging image based on the path source matrix and the cumulative cost matrix; Obtaining a feature preservation mask for the pixel point based on the row coordinates of the pixel point and the optimal row number of a column of pixel points matching the pixel point; The calculation formula of the feature preservation mask of the pixel point is as follows: ； in, Pixel The feature preservation mask of For pixels Matching The optimal row number of the column pixel, is the row difference threshold.

6. The crack detection method for a ribbed plate-shaped part according to claim 5, characterized in that: The calculation formula for the optimal row number of each column of pixels is as follows: ； in, The last column of the laser scanning thermal image The optimal row number of the pixel, is the maximum value of the row coordinate of the pixel point, is the maximum value of the column coordinates of the pixel point, For the The optimal row number of the column pixel, is the row coordinate of the minimum cumulative gradient magnitude in the last column of the cumulative cost matrix.

7. The crack detection method for a ribbed plate-shaped part according to claim 1, characterized in that: The step of obtaining a laser scanning thermal image of a ribbed plate-shaped component includes: Scan the uniformly rotating ribbed plate using a laser scanning thermal imaging system to obtain multiple original local laser scanning thermal images; reconstructing an original global laser scanning thermal image of the ribbed plate-shaped component based on the plurality of original local laser scanning thermal images; Anisotropic diffusion filtering is performed on the original global laser scanning thermal imaging image to obtain the laser scanning thermal imaging image.

8. A crack detection device for a ribbed plate-shaped part, characterized in that: include: An acquisition module, used for acquiring a laser scanning thermal image of a ribbed plate-shaped part; A first processing module is used to obtain a cost matrix of gradient amplitudes of pixels in the laser scanning thermal imaging image; A second processing module, configured to construct a path source matrix of the pixel points based on the cost matrix; a third processing module, configured to obtain a feature-preserving mask of the pixel points based on the path source matrix to obtain a feature map of the laser scanning thermal imaging image; An analysis module is used to analyze cracks in the ribbed plate-shaped component based on the characteristic graph.

9. An electronic device comprising a memory, a processor, and a computer program stored in the memory and running on the processor, characterized in that: When the processor executes the computer program, the crack detection method for the ribbed plate member according to any one of claims 1 to 7 is implemented.

10. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the crack detection method for a ribbed plate member according to any one of claims 1 to 7 is implemented.

Images