Engine piston surface defect detection method and system based on deep learning

Abstract

The invention discloses an engine piston surface defect detection method and system based on deep learning, and relates to the technical field of image processing, and the method comprises the steps: obtaining a standard image of an engine piston, and carrying out the position analysis and parameter analysis of the standard image of the engine piston, and determining position coordinates of the engine piston and shooting parameters of the engine piston. According to the method, pixel value change analysis is carried out on all images in a standard image set of the engine piston, a pixel value reference curve of the standard engine piston is determined, then pixel value change analysis is carried out on a to-be-analyzed image of the engine piston, a pixel value change curve of the to-be-analyzed image is determined, and finally, a pixel value reference curve of the to-be-analyzed image is determined. Whether the engine piston to be detected has defects or not can be determined by judging whether the pixel value change curve of the image to be analyzed is in the pixel value reference curve of the standard engine piston or not, the defects of the engine piston can be accurately detected, and the working intensity of workers can be reduced.

Description

Technical Field

[0001] This invention relates to the field of image processing technology, specifically to a method and system for detecting surface defects on engine pistons based on deep learning. Background Technology

[0002] The engine piston is the "heart" of a fuel engine. It plays a crucial role in converting the chemical energy of fuel into mechanical energy under extremely harsh working conditions.

[0003] Currently, defect detection of engine pistons is carried out manually. When workers conduct defect detection for a long time, misjudgments may occur, which will indirectly reduce the production quality of the engine. Summary of the Invention

[0004] To address the aforementioned technical problems, a method and system for detecting surface defects on engine pistons based on deep learning is provided. This technical solution solves the problem mentioned in the background art that the existing defect detection of engine pistons is carried out manually. When workers perform defect detection for a long time, misjudgments may occur, which indirectly reduces the production quality of the engine.

[0005] To achieve the above objectives, the technical solution adopted by the present invention is as follows:

[0006] A deep learning-based method for detecting surface defects in engine pistons, comprising:

[0007] A standard image of the engine piston is acquired, and positional and parameter analyses are performed on the standard image of the engine piston to determine the position coordinates of the engine piston and the imaging parameters of the engine piston; wherein, the imaging parameters of the engine piston include the imaging background of the engine piston and the parameter information of the standard image.

[0008] Based on the location coordinates of the engine piston and the imaging parameters of the engine piston, the engine piston to be detected is image captured and processed to obtain the image of the engine piston to be analyzed.

[0009] Based on standard images of engine pistons, comparative analysis is performed on the images of engine pistons to be analyzed to determine whether there are defects in the engine pistons to be inspected.

[0010] Preferably, the steps of acquiring a standard image of the engine piston, performing position and parameter analysis on the standard image of the engine piston, and determining the position coordinates of the engine piston and the imaging parameters of the engine piston specifically include the following steps:

[0011] Data extraction and processing are performed on the database system to obtain a standard image set of engine pistons; wherein, the background of the engine piston and the parameter information of the standard images in the standard image set of engine pistons are completely identical.

[0012] Arbitrarily select a standard image of an engine piston from the set of standard images of engine pistons, perform data reading and processing on the standard image of the engine piston, and obtain the shooting background and parameter information of the standard image of the engine piston.

[0013] Construct a rectangular coordinate system based on the standard image of the engine piston, with the lower left corner of the standard image of the engine piston as the origin, the horizontal direction of the standard image of the engine piston as the X-axis, and the vertical direction of the standard image of the engine piston as the Y-axis.

[0014] Based on the Cartesian coordinate system, coordinate positioning processing is performed on the standard image of the engine piston to determine the coordinates of the engine piston's position.

[0015] Preferably, the step of performing image capture processing on the engine piston to be detected based on the engine piston's location coordinates and the engine piston's imaging parameters to obtain the engine piston image to be analyzed specifically includes the following steps:

[0016] The parameters of the imaging equipment for the engine piston are adjusted based on the parameter information of the standard image;

[0017] Based on the coordinates of the engine piston's location, the background of the engine piston image is positioned to determine the placement outline of the engine piston.

[0018] The placement contour of the engine piston and the position of the engine piston to be tested are matched, and the engine piston to be tested is placed into the placement contour of the engine piston.

[0019] The imaging device based on the engine piston performs image acquisition and processing on the engine piston to be inspected, and obtains the image of the engine piston to be analyzed.

[0020] Preferably, the step of comparing and analyzing the engine piston image to be analyzed based on a standard image of the engine piston to determine whether the engine piston to be inspected has defects specifically includes the following steps:

[0021] Perform coordinate analysis on a standard image of the engine piston to determine the pixel coordinates of the engine piston's edge contour.

[0022] Based on the pixel coordinates of the edge contour of the engine piston, data analysis and processing are performed on all images in the standard image set of the engine piston to determine the pixel value reference curve of the standard engine piston.

[0023] Based on the edge contour pixel coordinates of the engine piston and the pixel value reference curve of the standard engine piston, the image of the engine piston to be analyzed is subjected to curve comparison processing to determine whether there is a defect in the engine piston to be detected.

[0024] Preferably, the step of performing data analysis and processing on all images in the standard image set of the engine piston based on the pixel coordinates of the engine piston's edge contour to determine the pixel value reference curve of the standard engine piston specifically includes the following steps:

[0025] Based on the pixel coordinates of the edge contour of the engine piston, coordinate annotation is performed on all images in the standard image set of the engine piston to determine the coordinate position of the first annotated pixel value.

[0026] Based on the first labeled pixel value coordinate position, coordinate extraction processing is performed on all images in the standard image set of the engine piston to obtain the pixel value coordinate set of the engine piston; wherein, the number of data in the pixel value coordinate set of the engine piston is consistent with the number of images in the standard image set of the engine piston;

[0027] Pixel value analysis is performed on the set of pixel value coordinates related to the engine piston to determine the pixel value reference curve for a standard engine piston.

[0028] Preferably, the step of performing pixel value analysis on the set of pixel value coordinates related to the engine piston to determine the pixel value reference curve of the standard engine piston specifically includes the following steps:

[0029] Feature analysis is performed on the set of pixel value coordinates related to the engine piston to determine the set of pixel value matching functions for the standard image; wherein the number of functions in the set of pixel value matching functions is consistent with the number of data in the set of pixel value coordinates related to the engine piston.

[0030] Curve plotting is performed on all functions in the set of pixel value matching functions for the standard image to obtain several sets of pixel value change curves;

[0031] By comparing and analyzing several sets of pixel value variation curves, a reference curve for the pixel value of a standard engine piston is determined.

[0032] Preferably, the step of comparing and analyzing several sets of pixel value change curves to determine the pixel value reference curve of a standard engine piston specifically includes the following steps:

[0033] Compare several sets of pixel value change curves to determine the common and different parts of the pixel value change curves;

[0034] Based on the common part of the pixel value change curve, integrate this part of several sets of pixel value change curves into a single curve to obtain the integrated curve of pixel value change curve.

[0035] Based on different parts of the pixel value change curve, the relevant parts of several sets of pixel value change curves are truncated to obtain the difference curves of several sets of pixel value change curves.

[0036] By placing the integrated curve of pixel value variation curves and the difference curves of several sets of pixel value variation curves in the same coordinate system and performing curve connection processing, a pixel value reference curve for a standard engine piston is obtained.

[0037] Preferably, the step of performing curve comparison processing on the engine piston image to be analyzed based on the edge contour pixel coordinates of the engine piston and the pixel value reference curve of a standard engine piston to determine whether the engine piston to be detected has defects specifically includes the following steps:

[0038] Based on the pixel coordinates of the edge contour of the engine piston, the image of the engine piston to be analyzed is subjected to coordinate annotation processing to determine the coordinate position of the second annotation pixel value.

[0039] Based on the coordinate position of the second labeled pixel value, the coordinate extraction process is performed on the image of the engine piston to be analyzed to obtain the pixel value coordinates of the engine piston in the image to be analyzed.

[0040] Feature analysis is performed on the pixel coordinates of the engine piston in the image to be analyzed to determine the pixel value matching function of the image to be analyzed.

[0041] The pixel value matching function of the image to be analyzed is used to perform curve plotting to obtain the pixel value change curve of the image to be analyzed;

[0042] Based on the pixel value reference curve of a standard engine piston, a comparative analysis is performed on the pixel value change curve of the image to be analyzed to determine whether there is a defect in the engine piston to be inspected.

[0043] Preferably, the step of comparing and analyzing the pixel value change curve of the image to be analyzed based on the pixel value reference curve of a standard engine piston to determine whether the engine piston to be detected has defects specifically includes the following steps:

[0044] The pixel value reference curve of the standard engine piston and the pixel value change curve of the image to be analyzed are matched.

[0045] If the pixel value change curve of the image to be analyzed is inside the pixel value reference curve of the standard engine piston, the engine piston to be tested has no defects.

[0046] If the pixel value change curve of the image to be analyzed is not completely within the pixel value reference curve of the standard engine piston, the engine piston to be tested has a defect.

[0047] Furthermore, a deep learning-based engine piston surface defect detection system is proposed to implement the deep learning-based engine piston surface defect detection method described above, including:

[0048] The intelligent analysis terminal is used to control various modules to compare and analyze the standard image of the engine piston and the image of the engine piston to be analyzed, and to determine whether there is a defect in the engine piston to be detected.

[0049] A database system for storing a standard set of images of engine pistons;

[0050] The parameter determination module is used to extract parameters from images in a standard image set of engine pistons to determine the coordinates of the engine piston's location and the imaging parameters of the engine piston.

[0051] The imaging device is used to acquire and process images of the engine piston to be inspected, and to obtain an image of the engine piston to be analyzed.

[0052] The first function matching module performs function matching processing based on the set of pixel value coordinates of the engine piston to determine the set of pixel value matching functions for the standard image.

[0053] The curve analysis module is used to perform curve analysis processing on the set of pixel value matching functions of the standard image to determine the pixel value reference curve of the standard engine piston.

[0054] The second function matching module performs function matching processing based on the pixel value coordinates of the engine piston in the image to be analyzed, and determines the pixel value matching function of the image to be analyzed.

[0055] The defect analysis module performs matching processing on the pixel value matching function of the image to be analyzed based on the pixel value reference curve of the standard engine piston to determine whether there is a defect in the engine piston to be detected.

[0056] Compared with existing technologies, this invention provides a method and system for detecting surface defects on engine pistons based on deep learning, which has the following advantages:

[0057] This invention analyzes pixel value changes in all images within a standard image set of engine pistons to determine a reference curve for standard engine piston pixel values. Then, it analyzes pixel value changes in the image of the engine piston to be analyzed to determine the pixel value change curve of the image to be analyzed. Finally, it determines whether the pixel value change curve of the image to be analyzed falls within the reference curve for standard engine piston pixel values, thus identifying whether the engine piston to be inspected has defects. This method not only accurately detects engine piston defects but also reduces the workload of operators. Attached Figure Description

[0058] Figure 1 This is a flowchart illustrating steps S100-S300 in a deep learning-based engine piston surface defect detection method proposed in this invention.

[0059] Figure 2 This is a structural block diagram of an engine piston surface defect detection system based on deep learning proposed in this invention. Detailed Implementation

[0060] The following description is intended to disclose the invention and enable those skilled in the art to implement it. The preferred embodiments described below are merely examples, and other obvious variations will occur to those skilled in the art.

[0061] Reference Figure 1 As shown, a deep learning-based method for detecting surface defects in engine pistons includes:

[0062] S100. Obtain a standard image of the engine piston, perform position and parameter analysis on the standard image of the engine piston, and determine the position coordinates of the engine piston and the shooting parameters of the engine piston; wherein, the shooting parameters of the engine piston include the shooting background of the engine piston and the parameter information of the standard image.

[0063] S200: Based on the coordinates of the engine piston's location and the engine piston's imaging parameters, perform image capture processing on the engine piston to be detected to obtain the image of the engine piston to be analyzed.

[0064] S300: Based on the standard image of the engine piston, compare and analyze the image of the engine piston to be analyzed to determine whether there is a defect in the engine piston to be inspected;

[0065] Those skilled in the art will understand that, due to the influence of the manufacturing process, the yield rate of engine pistons cannot be 100%, and there will be a certain number of defective products. However, these defective products will be mixed with good products. If the defective products are not screened out and put directly into use, the quality of the engine using the defective pistons will be reduced, or even rendered unusable, thereby increasing the engine's production cost. If the produced engine pistons are inspected manually, it will not only increase the workload of the workers, but also may lead to misjudgments. When misjudgments occur, good products may be mixed with defective products, and defective products may be mixed with good products. Both of these situations will increase the engine's production cost. Therefore, by comparing and analyzing the standard image of the engine piston with the image to be analyzed, it is possible to determine whether there are defects in the engine piston.

[0066] Example 1

[0067] Step S100: Obtain a standard image of the engine piston, perform position and parameter analysis on the standard image of the engine piston, and determine the position coordinates of the engine piston and the imaging parameters of the engine piston. Specifically, this includes the following steps:

[0068] S101. Perform data extraction processing on the database system to obtain a standard image set of engine pistons; wherein, the shooting background of the engine pistons and the parameter settings of the shooting equipment in the standard image set of engine pistons are completely identical.

[0069] S102. Select any standard image of an engine piston from the set of standard images of engine pistons, perform data reading and processing on the standard image of the engine piston, and obtain the shooting background and parameter information of the standard image of the engine piston.

[0070] S103. Construct a rectangular coordinate system based on the standard image of the engine piston, with the lower left corner of the standard image of the engine piston as the origin, the horizontal direction of the standard image of the engine piston as the X-axis, and the vertical direction of the standard image of the engine piston as the Y-axis.

[0071] S104. Based on the rectangular coordinate system, perform coordinate positioning processing on the standard image of the engine piston to determine the coordinates of the engine piston's position.

[0072] In this embodiment, if the engine piston to be detected has a defect, some pixel values ​​in the captured image of the engine piston to be detected will be different from the pixel values ​​in the image of a normal engine piston. However, the image of the engine piston to be detected and the image of a normal engine piston need to be captured under the same shooting background and with the same parameter information as the standard image. Furthermore, the position of the engine piston to be detected in the shooting background also needs to be consistent with its position in the standard image of the engine piston. Therefore, by constructing a Cartesian coordinate system, the shooting position of the engine piston in the standard image of the engine piston (i.e., the coordinates of the engine piston's position) is determined. Here, the coordinates of the engine piston's position mainly refer to the edge contour of the engine piston. Subsequently, the engine piston to be detected can be image acquired simply by placing it within the edge contour of the engine piston.

[0073] Example 2

[0074] Step S200: Based on the position coordinates of the engine piston and the imaging parameters of the engine piston, image processing is performed on the engine piston to be detected to obtain the image of the engine piston to be analyzed. This specifically includes the following steps:

[0075] S201. Adjust the parameters of the imaging device for the engine piston according to the parameter information of the standard image;

[0076] S202. Based on the coordinates of the engine piston's location, perform positioning processing on the background of the engine piston's image to determine the placement outline of the engine piston.

[0077] S203. Match the position of the engine piston placement contour with the engine piston to be tested, and place the engine piston to be tested into the engine piston placement contour.

[0078] S204. An image acquisition device based on an engine piston is used to acquire and process images of the engine piston to be inspected, and to obtain an image of the engine piston to be analyzed.

[0079] In this embodiment, if the placement position of the engine piston to be detected is different from the coordinates of the engine piston's location, it is impossible to compare the standard image of the engine piston and the image of the engine piston to be analyzed. Because the placement position is different, the shooting background may change, which will cause the pixel value of the captured image to change. If the comparison continues, misjudgment may occur. Therefore, before taking an image of the engine piston to be detected, it is necessary to determine the placement position of the engine piston to be detected.

[0080] Example 3

[0081] Step S300: Based on the standard image of the engine piston, compare and analyze the image of the engine piston to be analyzed to determine whether there is a defect in the engine piston to be inspected. This specifically includes the following steps:

[0082] S301. Perform coordinate analysis on the standard image of the engine piston to determine the pixel coordinates of the edge contour of the engine piston.

[0083] S302. Based on the pixel coordinates of the edge contour of the engine piston, perform data analysis and processing on all images in the standard image set of the engine piston to determine the pixel value reference curve of the standard engine piston.

[0084] S303. Based on the edge contour pixel coordinates of the engine piston and the pixel value reference curve of the standard engine piston, perform curve comparison processing on the image of the engine piston to be analyzed to determine whether there is a defect in the engine piston to be detected.

[0085] Understandably, a standard image of an engine piston is a single image. If defect analysis is performed based on a single image, the false positive rate may increase because the standard image of the engine piston may have different pixel values ​​due to varying lighting intensities. To avoid false positives, multiple standard images of the engine piston are analyzed to determine the pixel values ​​of a normal engine piston image under different lighting conditions.

[0086] Specifically, step S302, which involves performing data analysis and processing on all images in the standard image set of the engine piston based on the pixel coordinates of the engine piston's edge contour, to determine the pixel value reference curve for the standard engine piston, includes the following steps:

[0087] S3021. Based on the pixel coordinates of the edge contour of the engine piston, perform coordinate annotation processing on all images in the standard image set of the engine piston to determine the coordinate position of the first annotated pixel value.

[0088] The pixel coordinates of the edge contour of the engine piston can be determined using algorithms such as YOLOv8, Faster R-CNN, and U-Net.

[0089] S3022. Based on the coordinate position of the first labeled pixel value, perform coordinate extraction processing on all images in the standard image set of the engine piston to obtain a set of pixel value coordinates about the engine piston; wherein, the number of data in the set of pixel value coordinates about the engine piston is the same as the number of images in the standard image set of the engine piston.

[0090] S3023. Perform pixel value analysis on the set of pixel value coordinates for the engine piston to determine the pixel value reference curve for the standard engine piston.

[0091] Understandably, in order to reduce the amount of subsequent calculations, the pixel values ​​in the standard image of the engine piston are filtered out to identify the pixel values ​​belonging to the engine piston. Therefore, the position of the engine piston in the image is determined by the pixel coordinates of the edge contour of the engine piston. Subsequently, all pixel values ​​belonging to the engine piston can be determined simply by using the pixel coordinates of the edge contour of the engine piston.

[0092] Specifically, step S3023, analyzing the pixel value coordinates of the engine piston to determine the standard engine piston pixel value reference curve, includes the following steps:

[0093] S30231. Perform feature analysis processing on the set of pixel value coordinates related to the engine piston to determine the set of pixel value matching functions for the standard image; wherein, the number of functions in the set of pixel value matching functions is consistent with the number of data in the set of pixel value coordinates related to the engine piston;

[0094] S30232. Perform curve plotting on all functions in the set of pixel value matching functions for the standard image to obtain several sets of pixel value change curves;

[0095] S30233. Compare and analyze several sets of pixel value change curves to determine the pixel value reference curve for a standard engine piston.

[0096] It is understandable that the pixel values ​​of an image can be replaced by a function, that is, the image can be expressed by a function expression, and the image can also be expressed by a curve. Therefore, by converting the pixel values ​​of the engine piston into a function, the function can then be plotted as a curve.

[0097] Specifically, step S30233, comparing and analyzing several sets of pixel value change curves to determine the pixel value reference curve for a standard engine piston, includes the following steps:

[0098] S302331. Compare several sets of pixel value change curves to determine the common and different parts of the pixel value change curves;

[0099] S302332. Based on the common part of the pixel value change curve, integrate the same part of several sets of pixel value change curves into a single curve to obtain the integrated curve of pixel value change curve.

[0100] S302333: Based on different parts of the pixel value change curve, the relevant parts of several sets of pixel value change curves are truncated to obtain the difference curves of several sets of pixel value change curves.

[0101] S302334. Place the integrated curve of the pixel value change curve and the difference curve of several sets of pixel value change curves in the same coordinate system and perform curve connection processing to obtain the pixel value reference curve of the standard engine piston.

[0102] It is understandable that due to different light intensities, some pixel values ​​of the standard image of the engine piston are different. However, these pixel values ​​are also the pixel values ​​of a normal engine piston. In order to avoid multiple comparisons, these curves are integrated. The curves representing the same pixel values ​​are replaced by a single curve, while the curves representing different pixel values ​​remain unchanged. However, the starting and ending points of this part of the curve are connected to the curves representing the same pixel values, which is the reference curve for the pixel values ​​of the standard engine piston.

[0103] Specifically, step S303, which involves comparing the edge contour pixel coordinates of the engine piston with the pixel value reference curve of a standard engine piston to determine whether the engine piston to be detected has defects, includes the following steps:

[0104] S3031. Based on the pixel coordinates of the edge contour of the engine piston, perform coordinate annotation processing on the image of the engine piston to be analyzed, and determine the coordinate position of the second annotation pixel value.

[0105] S3032. Based on the coordinate position of the second labeled pixel value, perform coordinate extraction processing on the image of the engine piston to be analyzed to obtain the pixel value coordinates of the engine piston in the image to be analyzed.

[0106] S3033. Perform feature analysis processing on the pixel value coordinates of the engine piston in the image to be analyzed, and determine the pixel value matching function of the image to be analyzed.

[0107] S3034. Perform curve plotting on the pixel value matching function of the image to be analyzed to obtain the pixel value change curve of the image to be analyzed;

[0108] S3035. Based on the pixel value reference curve of a standard engine piston, perform comparative analysis on the pixel value change curve of the image to be analyzed to determine whether the engine piston to be detected has defects.

[0109] Specifically, step S3035, which involves comparing and analyzing the pixel value change curve of the image to be analyzed based on the pixel value reference curve of a standard engine piston to determine whether the engine piston to be detected has defects, includes the following steps:

[0110] S30351. Match the pixel value reference curve of the standard engine piston with the pixel value change curve of the image to be analyzed;

[0111] S30352. If the pixel value change curve of the image to be analyzed is inside the pixel value reference curve of the standard engine piston, the engine piston to be tested has no defects.

[0112] S30353. If the pixel value change curve of the image to be analyzed is not completely within the pixel value reference curve of the standard engine piston, the engine piston to be tested has a defect.

[0113] It is understandable that when the image of the engine piston to be analyzed is free of defects, the pixel value change curve corresponding to the image of the engine piston to be analyzed should be inside the pixel value reference curve of the standard engine piston. However, when the image of the engine piston to be analyzed is defective, part of the pixel value change curve corresponding to the image of the engine piston to be analyzed should be outside the pixel value reference curve of the standard engine piston.

[0114] Reference Figure 2 As shown, a deep learning-based engine piston surface defect detection system is used to implement the deep learning-based engine piston surface defect detection method described above, including:

[0115] The intelligent analysis terminal is used to control various modules to compare and analyze the standard image of the engine piston and the image of the engine piston to be analyzed, and to determine whether there is a defect in the engine piston to be detected.

[0116] A database system for storing a standard set of images of engine pistons;

[0117] The parameter determination module is used to extract parameters from images in a standard image set of engine pistons to determine the coordinates of the engine piston's location and the imaging parameters of the engine piston.

[0118] The imaging device is used to acquire and process images of the engine piston to be inspected, and to obtain an image of the engine piston to be analyzed.

[0119] The first function matching module performs function matching processing based on the set of pixel value coordinates of the engine piston to determine the set of pixel value matching functions for the standard image.

[0120] The curve analysis module is used to perform curve analysis processing on the set of pixel value matching functions of the standard image to determine the pixel value reference curve of the standard engine piston.

[0121] The second function matching module performs function matching processing based on the pixel value coordinates of the engine piston in the image to be analyzed, and determines the pixel value matching function of the image to be analyzed.

[0122] The defect analysis module performs matching processing on the pixel value matching function of the image to be analyzed based on the pixel value reference curve of the standard engine piston to determine whether there is a defect in the engine piston to be detected.

[0123] The foregoing has shown and described the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The embodiments and descriptions in the specification are merely principles of the invention. Various changes and modifications can be made to the invention without departing from its spirit and scope, and all such changes and modifications fall within the scope of the claimed invention. The scope of protection claimed by the appended claims and their equivalents is defined.

Claims

1. A method for detecting surface defects on engine pistons based on deep learning, characterized in that, include: A standard image of the engine piston is acquired, and positional and parameter analyses are performed on the standard image of the engine piston to determine the position coordinates of the engine piston and the imaging parameters of the engine piston; wherein, the imaging parameters of the engine piston include the imaging background of the engine piston and the parameter information of the standard image. Based on the location coordinates of the engine piston and the imaging parameters of the engine piston, the engine piston to be detected is image captured and processed to obtain the image of the engine piston to be analyzed. Based on standard images of engine pistons, comparative analysis is performed on the images of engine pistons to be analyzed to determine whether there are defects in the engine pistons to be inspected.

2. The method for detecting surface defects of engine pistons based on deep learning according to claim 1, characterized in that, The process of acquiring a standard image of the engine piston, performing position and parameter analysis on the standard image of the engine piston, and determining the position coordinates of the engine piston and the imaging parameters of the engine piston specifically includes the following steps: Data extraction and processing are performed on the database system to obtain a standard image set of engine pistons; wherein the background of the engine piston and the parameter settings of the shooting equipment are exactly the same in the standard image set of engine pistons. Arbitrarily select a standard image of an engine piston from the set of standard images of engine pistons, perform data reading and processing on the standard image of the engine piston, and obtain the shooting background and parameter information of the standard image of the engine piston. Construct a rectangular coordinate system based on the standard image of the engine piston, with the lower left corner of the standard image of the engine piston as the origin, the horizontal direction of the standard image of the engine piston as the X-axis, and the vertical direction of the standard image of the engine piston as the Y-axis. Based on the Cartesian coordinate system, coordinate positioning processing is performed on the standard image of the engine piston to determine the coordinates of the engine piston's position.

3. The method for detecting surface defects of engine pistons based on deep learning according to claim 2, characterized in that, The process of capturing and processing images of the engine piston to be analyzed, based on the engine piston's location coordinates and imaging parameters, specifically includes the following steps: The parameters of the imaging equipment for the engine piston are adjusted based on the parameter information of the standard image; Based on the coordinates of the engine piston's location, the background of the engine piston image is positioned to determine the placement outline of the engine piston. The placement contour of the engine piston and the position of the engine piston to be tested are matched, and the engine piston to be tested is placed into the placement contour of the engine piston. The imaging device based on the engine piston performs image acquisition and processing on the engine piston to be inspected, and obtains the image of the engine piston to be analyzed.

4. The method for detecting surface defects of engine pistons based on deep learning according to claim 3, characterized in that, The process of comparing and analyzing the standard image of the engine piston with the image to be analyzed to determine whether the engine piston to be inspected has defects includes the following steps: Perform coordinate analysis on a standard image of the engine piston to determine the pixel coordinates of the engine piston's edge contour. Based on the pixel coordinates of the edge contour of the engine piston, data analysis and processing are performed on all images in the standard image set of the engine piston to determine the pixel value reference curve of the standard engine piston. Based on the edge contour pixel coordinates of the engine piston and the pixel value reference curve of the standard engine piston, the image of the engine piston to be analyzed is subjected to curve comparison processing to determine whether there is a defect in the engine piston to be detected.

5. The method for detecting surface defects of engine pistons based on deep learning according to claim 4, characterized in that, The step of performing data analysis and processing on all images in the standard image set of the engine piston based on the pixel coordinates of the engine piston's edge contour to determine the pixel value reference curve of the standard engine piston specifically includes the following steps: Based on the pixel coordinates of the edge contour of the engine piston, coordinate annotation is performed on all images in the standard image set of the engine piston to determine the coordinate position of the first annotated pixel value. Based on the first labeled pixel value coordinate position, coordinate extraction processing is performed on all images in the standard image set of the engine piston to obtain the pixel value coordinate set of the engine piston; wherein, the number of data in the pixel value coordinate set of the engine piston is the same as the number of images in the standard image set of the engine piston; Pixel value analysis is performed on the set of pixel value coordinates related to the engine piston to determine the pixel value reference curve for a standard engine piston.

6. The method for detecting surface defects of engine pistons based on deep learning according to claim 5, characterized in that, The step of performing pixel value analysis on the set of pixel value coordinates related to the engine piston to determine the pixel value reference curve for a standard engine piston specifically includes the following steps: Feature analysis is performed on the set of pixel value coordinates related to the engine piston to determine the set of pixel value matching functions for the standard image; wherein the number of functions in the set of pixel value matching functions is consistent with the number of data in the set of pixel value coordinates related to the engine piston. Curve plotting is performed on all functions in the set of pixel value matching functions for the standard image to obtain several sets of pixel value change curves; By comparing and analyzing several sets of pixel value variation curves, a reference curve for the pixel value of a standard engine piston is determined.

7. The method for detecting surface defects of engine pistons based on deep learning according to claim 6, characterized in that, The process of comparing and analyzing several sets of pixel value variation curves to determine the pixel value reference curve for a standard engine piston specifically includes the following steps: Compare several sets of pixel value change curves to determine the common and different parts of the pixel value change curves; Based on the common part of the pixel value change curve, integrate this part of several sets of pixel value change curves into a single curve to obtain the integrated curve of pixel value change curve. Based on different parts of the pixel value change curve, the relevant parts of several sets of pixel value change curves are truncated to obtain the difference curves of several sets of pixel value change curves. By placing the integrated curve of pixel value variation curves and the difference curves of several sets of pixel value variation curves in the same coordinate system and performing curve connection processing, a pixel value reference curve for a standard engine piston is obtained.

8. The method for detecting surface defects of engine pistons based on deep learning according to claim 7, characterized in that, The process of comparing the edge contour pixel coordinates of the engine piston with the pixel value reference curve of a standard engine piston to determine whether there are defects in the engine piston to be detected includes the following steps: Based on the pixel coordinates of the edge contour of the engine piston, the image of the engine piston to be analyzed is subjected to coordinate annotation processing to determine the coordinate position of the second annotation pixel value. Based on the coordinate position of the second labeled pixel value, the coordinate extraction process is performed on the image of the engine piston to be analyzed to obtain the pixel value coordinates of the engine piston in the image to be analyzed. Feature analysis is performed on the pixel coordinates of the engine piston in the image to be analyzed to determine the pixel value matching function of the image to be analyzed. The pixel value matching function of the image to be analyzed is used to perform curve plotting to obtain the pixel value change curve of the image to be analyzed; Based on the pixel value reference curve of a standard engine piston, a comparative analysis is performed on the pixel value change curve of the image to be analyzed to determine whether there is a defect in the engine piston to be inspected.

9. The method for detecting surface defects of engine pistons based on deep learning according to claim 8, characterized in that, The process of comparing and analyzing the pixel value variation curve of the image to be analyzed based on the pixel value reference curve of a standard engine piston to determine whether there is a defect in the engine piston to be detected includes the following steps: The pixel value reference curve of the standard engine piston and the pixel value change curve of the image to be analyzed are matched. If the pixel value change curve of the image to be analyzed is inside the pixel value reference curve of the standard engine piston, the engine piston to be tested has no defects. If the pixel value change curve of the image to be analyzed is not completely within the pixel value reference curve of the standard engine piston, the engine piston to be tested has a defect.

10. A deep learning-based engine piston surface defect detection system, used to implement the deep learning-based engine piston surface defect detection method as described in any one of claims 1-9, characterized in that, include: The intelligent analysis terminal is used to control various modules to compare and analyze the standard image of the engine piston and the image of the engine piston to be analyzed, and to determine whether there is a defect in the engine piston to be detected. A database system for storing a standard set of images of engine pistons; The parameter determination module is used to extract parameters from images in a standard image set of engine pistons to determine the coordinates of the engine piston's location and the imaging parameters of the engine piston; the imaging device is used to acquire and process images of the engine piston to be detected to obtain the image of the engine piston to be analyzed. The first function matching module performs function matching processing based on the set of pixel value coordinates of the engine piston to determine the set of pixel value matching functions for the standard image. The curve analysis module is used to perform curve analysis processing on the set of pixel value matching functions of the standard image to determine the pixel value reference curve of the standard engine piston. The second function matching module performs function matching processing based on the pixel value coordinates of the engine piston in the image to be analyzed, and determines the pixel value matching function of the image to be analyzed. The defect analysis module performs matching processing on the pixel value matching function of the image to be analyzed based on the pixel value reference curve of the standard engine piston to determine whether there is a defect in the engine piston to be detected.

Images