Automated monitoring system and method for flat blade patterns of automotive parts

Abstract

The application belongs to the technical field of detection, and provides an automatic monitoring system and method for planar knife lines of automobile parts. The method comprises the following steps: inputting a region knife line feature set obtained from a high-definition image into a classifier, integrating a plurality of knife line probabilities output by the classifier to obtain a first knife line probability corresponding to the region knife line feature set; if the first knife line probability is lower than a first probability and higher than a second probability, inputting the region knife line feature set, a plurality of other region knife line feature sets determined to have planar knife lines, and function attribute information of the automobile part into a deep analysis model; if a second knife line probability output by the deep analysis model is higher than the first probability, determining that a region corresponding to the region knife line feature set of the automobile part has a planar knife line; and labeling the automobile part region determined to have a planar knife line. The application can realize efficient quality inspection and labeling of planar knife lines on automobile parts, and assist in subsequent processing of the planar knife lines.

Description

Technical Field

[0001] This invention relates to the field of detection technology, and more specifically, to an automatic monitoring system and method for detecting planar tool marks on automotive parts. Background Technology

[0002] In modern automotive manufacturing, the precision and quality of automotive parts play a crucial role in the overall performance and safety of vehicles. With the rapid development of the automotive industry, the requirements for the processing technology of automotive parts are also increasing.

[0003] Currently, during the processing of automotive parts, various factors often lead to tool marks on the surface of the parts, as shown in the circled area in Figure 1. On the one hand, tool wear, improper selection of cutting parameters, and machine tool precision issues can all prevent the tool from removing material stably and accurately during the cutting process, thus leaving tool marks on the surface of the parts. For example, after prolonged use, the cutting edge of the tool will gradually wear down, leading to unstable cutting forces and consequently tool marks. On the other hand, the properties of different automotive parts materials, such as aluminum alloys and cast iron, also affect the probability of tool marks forming. Some materials have uneven hardness, making them more prone to tool marks during processing.

[0004] For sealed automotive components, the problem of tool marks on the surface is particularly serious. The sealing performance of sealed automotive components, such as engine blocks and transmission housings, directly affects the normal operation of the vehicle. If tool marks exist on the surface of these components, a tight seal cannot be achieved after assembly, allowing gas or liquid to leak through the tiny gaps created by the tool marks. This leakage not only reduces vehicle performance, such as causing a decrease in engine power and transmission fluid leaks, but may also pose safety hazards, threatening driving safety.

[0005] How to automatically monitor the planar tool marks on automotive parts is a technical problem that needs to be solved. Summary of the Invention

[0006] To address the aforementioned technical problems, this invention provides an automatic monitoring method, system, electronic device, computer storage medium, and computer program product for planar tool marks on automotive parts.

[0007] This invention discloses an automatic monitoring method for planar tool marks on automotive parts. The method includes the following steps: under uniform light source illumination, high-definition images of the automotive parts are captured from multiple angles; tool mark features are extracted from each of the high-definition images to obtain multiple regional tool mark feature sets; wherein, each regional tool mark feature set contains multiple tool mark feature groups, and each tool mark feature group corresponds to a high-definition image at a certain angle; each tool mark feature group in the regional tool mark feature set is input into a classifier, and the multiple tool mark probabilities output by the classifier are integrated to obtain a first tool mark probability corresponding to the regional tool mark feature set; if the first tool mark probability is higher than the first probability, it is determined that the automotive part region corresponding to the regional tool mark feature set has planar tool marks; if the first tool mark probability is lower than the first probability but higher than the second probability, the regional tool mark feature set, several other regional tool mark feature sets that are determined to have planar tool marks, and the functional attribute information of the automotive part are input into a depth analysis model; if the second tool mark probability output by the depth analysis model is higher than the first probability, it is determined that the automotive part region corresponding to the regional tool mark feature set has planar tool marks; and the automotive part regions determined to have planar tool marks are tagged.

[0008] Optionally, the step of extracting knife-mark features from each of the high-definition images to obtain multiple regional knife-mark feature sets includes: image preprocessing: converting each of the acquired high-definition images to grayscale; removing image noise using Gaussian filtering or median filtering; enhancing image contrast through histogram equalization to highlight texture features; texture region segmentation: extracting texture contours using edge detection algorithms, connecting broken edges using morphological operations to form complete texture regions, and separating texture regions from the background using region growing algorithms or threshold segmentation methods; texture feature extraction: extracting the geometric features of the texture, including direction, spacing, depth, and shape; calculating the texture features of the texture; analyzing the frequency features of the texture using Fourier transform or wavelet transform; integrating the geometric features, texture features, and frequency features of the texture in a single region extracted from each of the high-definition images into the knife-mark feature group, and integrating each of the knife-mark feature groups into a regional knife-mark feature set corresponding to that single region.

[0009] Optionally, inputting the tool mark feature set of the region, several other region tool mark feature sets that are determined to have planar tool marks, and the functional attribute information of the automotive part into the depth analysis model includes: determining the location of the tool mark feature set of the region on the automotive part, and determining the target surface region of the automotive part based on the location; determining the tool mark feature sets of other regions that are determined to have planar tool marks in the target surface region as auxiliary region tool mark feature sets; inputting the tool mark feature set of the region, each of the auxiliary region tool mark feature sets, and the functional attribute information of the automotive part into the depth analysis model, and the depth analysis model outputting a third tool mark probability; wherein, the functional attribute information includes the application type and installation location of the automotive part; and calculating the second tool mark probability based on the third tool mark probability.

[0010] Optionally, calculating the second knife-mark probability based on the third knife-mark probability includes: extracting the accessory features of the automotive accessory from any of the high-definition images; determining whether the location belongs to the inner or outer side of the automotive accessory based on the accessory features; configuring a first coefficient if the location belongs to the inner side of the automotive accessory; configuring a second coefficient if the location belongs to the outer side of the automotive accessory; wherein the first coefficient is greater than the second coefficient; and optimizing the third knife-mark probability using the first coefficient or the second coefficient to obtain the second knife-mark probability.

[0011] Optionally, determining the target surface area of ​​the automotive parts based on the location includes: determining the probability difference between the first blade pattern probability and the first probability in the feature set of the blade pattern in the region; matching the surface radius according to the magnitude of the probability difference; wherein the surface radius is positively correlated with the probability difference; and determining the target surface area on the surface of the automotive parts based on the location and the surface radius.

[0012] This invention also discloses an automatic monitoring system for planar tool marks on automotive parts. The system includes a processing device and a storage device. Computer code stored in the storage device is called and executed by the processing device to achieve the following steps: under uniform light source illumination, high-definition images of automotive parts are captured from multiple angles; tool mark features are extracted from each of the high-definition images to obtain multiple regional tool mark feature sets; wherein, each regional tool mark feature set contains multiple tool mark feature groups, and each tool mark feature group corresponds to a high-definition image at a certain angle; each tool mark feature group in the regional tool mark feature set is input into a classifier, and the multiple tool mark probabilities output by the classifier are integrated. The system obtains the first blade pattern probability corresponding to the blade pattern feature set of the region; if the first blade pattern probability is higher than the first probability, it determines that the automotive part region corresponding to the blade pattern feature set of the region has planar blade patterns; if the first blade pattern probability is lower than the first probability but higher than the second probability, the system inputs the blade pattern feature set of the region, several other region blade pattern feature sets that are determined to have planar blade patterns, and the functional attribute information of the automotive part into the depth analysis model; if the second blade pattern probability output by the depth analysis model is higher than the first probability, it determines that the automotive part region corresponding to the blade pattern feature set of the region has planar blade patterns; and labels the automotive part regions that are determined to have planar blade patterns.

[0013] The present invention also discloses an electronic device comprising: at least one processor, a memory, and a computer program stored in the memory and executable on the at least one processor, the processor executing the computer program to implement the method as described in any of the preceding methods.

[0014] The present invention also discloses a computer storage medium storing a computer program that is executed by a processor to implement the methods described in any of the preceding methods.

[0015] The present invention also discloses a computer program product containing computer code, which, when executed by a processor of an electronic device, implements the method described in any of the preceding methods.

[0016] The beneficial effects of the present invention are at least as follows: 1) By acquiring high-definition images from multiple angles and extracting knife pattern features, the surface information of the parts can be fully captured, effectively reducing the omission of knife patterns due to shooting limitations and greatly improving the comprehensiveness of the detection.

[0017] 2) The probability judgment mechanism based on classifiers and deep analysis models significantly improves detection accuracy, reduces false judgment rate, and accurately identifies knife marks compared to traditional detection methods. Attached Figure Description

[0018] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the embodiments will be briefly introduced below. It should be understood that the following drawings only show some embodiments of the present invention and should not be regarded as a limitation on the scope. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0019] Figure 1 This is a schematic diagram of the planar tool marks present in the automotive parts disclosed in the embodiments of the present invention.

[0020] Figure 2 This is a flowchart illustrating an automatic monitoring method for planar tool marks on automotive parts, as disclosed in an embodiment of the present invention.

[0021] Figure 3 This is a schematic diagram of the structure of an automatic monitoring system for planar tool marks on automotive parts, as disclosed in an embodiment of the present invention. Detailed Implementation

[0022] The following specific embodiments illustrate the implementation of this application. Those skilled in the art can easily understand other advantages and effects of this application from the content disclosed in this specification. Obviously, the described embodiments are only some embodiments of this application, not all embodiments. Based on the embodiments in this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0023] Furthermore, the technical features involved in the different embodiments of this application described below can be combined with each other as long as they do not conflict with each other.

[0024] like Figure 2 As shown, in view of the above-mentioned technical problems, the present invention discloses an automatic monitoring method for planar tool marks on automotive parts. The method includes the following steps: S100, under the illumination of a uniform light source, high-definition images of automotive parts from multiple angles are captured, and tool mark features are extracted from each of the high-definition images to obtain multiple regional tool mark feature sets; wherein, the regional tool mark feature sets contain multiple tool mark feature groups, and each tool mark feature group corresponds to a high-definition image at one angle.

[0025] In this step, a uniform light source is set up in the assembly machine for automotive parts (such as a bushing assembly machine or a sealing machine) or in a specialized automotive parts quality inspection machine. Under the illumination of the uniform light source, a high-definition camera is controlled to capture high-definition images of the automotive parts from multiple angles.

[0026] For each high-resolution image, computer vision processing methods are used to identify regions that may contain planar knife marks, and knife mark features are extracted, thus obtaining multiple regions of knife mark feature sets. Each region of knife mark feature set corresponds to a region that may contain planar knife marks, and contains the knife mark feature set corresponding to that region extracted from each high-resolution image.

[0027] It should be noted that, due to the complex and diverse features of knife marks, extracting features from multiple angles can more comprehensively reflect the knife mark situation, and images from different angles are also conducive to highlighting the knife marks, which helps to improve the recognition accuracy of planar knife marks.

[0028] S200, each blade pattern feature group in the regional blade pattern feature set is input into the classifier, and the multiple blade pattern probabilities output by the classifier are integrated to obtain the first blade pattern probability corresponding to the regional blade pattern feature set; if the first blade pattern probability is higher than the first probability, it is determined that the automotive parts region corresponding to the regional blade pattern feature set has planar blade patterns.

[0029] In this step, the present invention constructs a classifier based on machine learning algorithms such as Support Vector Machine (SVM) and Random Forest. This classifier is small in size and has high processing efficiency. The classifier is used to perform preliminary and rapid classification and recognition of planar knife patterns in each knife pattern feature set of the region, obtaining multiple knife pattern probabilities. All knife pattern probabilities are then integrated, for example, by calculating the average value, to obtain the first knife pattern probability.

[0030] If the probability of the first knife mark is higher than the first probability, it means that after classifying the knife mark feature groups from different angles, the classifier determines that the existence of planar knife marks in the region is highly probable, that is, obvious knife marks. At this time, it is determined that planar knife marks exist in the region.

[0031] S300, if the first blade mark probability is lower than the first probability but higher than the second probability, then the blade mark feature set of this region, the blade mark feature sets of several other regions that are determined to have planar blade marks, and the functional attribute information of the automotive part are input into the depth analysis model. If the second blade mark probability output by the depth analysis model is higher than the first probability, then it is determined that the automotive part region corresponding to the blade mark feature set of this region has planar blade marks.

[0032] In this step, automotive parts may have certain actively designed textures, such as functional textures (e.g., anti-slip textures, heat dissipation grooves, air guides, etc.), decorative textures (e.g., brushed textures, matte textures, or other decorative textures), and surface treatment effects (sandblasting, electroplating, oxidation, and other surface treatment processes may create a knife-like effect on the part's surface). These actively designed textures have a certain visual similarity to planar knife marks, and it is difficult to determine the presence of knife marks based solely on the initial judgment of a classifier. Correspondingly, when the probability of a knife mark obtained based on the classifier is between the first probability and the second probability, it is identified as a suspicious knife mark, and this invention further uses a depth analysis model to further judge this area.

[0033] At this point, the feature sets of the knife-mark patterns in the suspicious area, the feature sets of the knife-mark patterns in other areas determined to contain planar knife-mark patterns, and the functional attribute information of the automotive parts are input into the deep analysis model. The deep analysis model combines this more comprehensive information for in-depth analysis and outputs a second knife-mark probability. If this probability is higher than the first probability, then the suspicious area is determined to indeed contain planar knife-mark patterns. This improves the accuracy of knife-mark detection, especially for cases that are difficult to determine directly; by integrating more information for in-depth analysis, the probability of missed detection is reduced.

[0034] S400 marks and labels areas of automotive parts that are determined to have planar knife marks.

[0035] In this step, areas with planar knife marks are labeled to facilitate targeted treatment of these areas later, such as manual verification and repair, to ensure the quality and safety of automotive parts and prevent problems such as seal leakage caused by knife marks from affecting the normal operation and safe driving of the vehicle.

[0036] The above-mentioned technical solution of the present invention has at least the following beneficial technical effects: 1) By acquiring high-definition images from multiple angles and extracting knife pattern features, the surface information of the parts is fully captured, effectively reducing the omission of knife patterns caused by shooting limitations and greatly improving the comprehensiveness of detection; 2) Based on the probability judgment mechanism of classifier and depth analysis model, compared with traditional detection methods, the detection accuracy is significantly improved, the false judgment rate is reduced, and the knife pattern is accurately identified.

[0037] Optionally, the step of extracting knife-mark features from each of the high-definition images to obtain multiple regional knife-mark feature sets includes: image preprocessing: converting each of the acquired high-definition images to grayscale; removing image noise using Gaussian filtering or median filtering; enhancing image contrast through histogram equalization to highlight texture features; texture region segmentation: extracting texture contours using edge detection algorithms, connecting broken edges using morphological operations to form complete texture regions, and separating texture regions from the background using region growing algorithms or threshold segmentation methods; texture feature extraction: extracting the geometric features of the texture, including direction, spacing, depth, and shape; calculating the texture features of the texture; analyzing the frequency features of the texture using Fourier transform or wavelet transform; integrating the geometric features, texture features, and frequency features of the texture in a single region extracted from each of the high-definition images into the knife-mark feature group, and integrating each of the knife-mark feature groups into a regional knife-mark feature set corresponding to that single region.

[0038] In this embodiment, the edge detection algorithm used to extract the texture contour is, for example, the Canny operator, and the morphological operations employed are, for example, dilation and erosion. Texture features of the texture include, for example, the contrast, energy, and entropy of the gray-level co-occurrence matrix (GLCM).

[0039] Optionally, inputting the tool mark feature set of the region, several other region tool mark feature sets that are determined to have planar tool marks, and the functional attribute information of the automotive part into the depth analysis model includes: determining the location of the tool mark feature set of the region on the automotive part, and determining the target surface region of the automotive part based on the location; determining the tool mark feature sets of other regions that are determined to have planar tool marks in the target surface region as auxiliary region tool mark feature sets; inputting the tool mark feature set of the region, each of the auxiliary region tool mark feature sets, and the functional attribute information of the automotive part into the depth analysis model, and the depth analysis model outputting a third tool mark probability; wherein, the functional attribute information includes the application type and installation location of the automotive part; and calculating the second tool mark probability based on the third tool mark probability.

[0040] In this embodiment, the appearance of tool marks on the surface of automotive parts is often not an isolated phenomenon; tool marks in a specific area may be related to surrounding areas. Therefore, this invention uses a depth analysis model to comprehensively analyze the tool mark feature set of the suspected tool mark area requiring secondary analysis, as well as the tool mark feature sets of other areas around it that have been identified as having planar tool marks. Furthermore, it uses the functional attribute information of the automotive part to comprehensively analyze whether the tool mark feature set of the area truly belongs to planar tool marks.

[0041] Auxiliary region tool mark feature sets can provide more surrounding information, helping deep analysis models to understand the situation in that region more comprehensively. For example, if a region is suspected of having tool marks, and similar texture features also exist in its surrounding areas, then the probability that the suspected tool mark is a real tool mark is high. Conversely, if there are no similar texture features in the surrounding areas, then the probability that the suspected tool mark is a real tool mark is lower. In other words, auxiliary region tool mark feature sets are valuable for determining whether the tool marks in that region are normal design patterns or machining-induced tool marks. However, the number of auxiliary region tool mark feature sets used should be limited, as excessive use will significantly reduce the analysis efficiency of deep analysis models. Moreover, if planar tool marks are caused by excessive wear of machine tool cutting tools, then planar tool marks usually exist in multiple regions. However, the appearance of planar tool marks is more related to the automotive parts themselves. For example, the special complex structure of automotive parts increases the difficulty of controlling the machining machine tool, causing the cutting tool of the machining machine tool to inevitably vibrate slightly, thus resulting in the existence of planar tool marks. In this case, the planar tool marks are actually regularly distributed in certain specific structural locations. Therefore, it is only necessary to use the feature set of the surrounding area corresponding to the area that needs to be analyzed in the second step as the auxiliary feature set of the area, without having to input the feature set of all the areas on the car part that are determined to have planar knife patterns into the depth analysis model.

[0042] After inputting the feature sets of the tool marks in the target area, the feature sets of the tool marks in each auxiliary area, and the functional attribute information of the automotive parts into the depth analysis model, the model can output a third tool mark probability. This third tool mark probability reflects the likelihood of tool marks existing in the target area based on the various input information. Finally, the second tool mark probability, as described above, is calculated from the third tool mark probability obtained from the depth analysis model.

[0043] It should be noted that the functional attribute information of automotive parts includes their application type (whether for sealing, transmission, or other functions) and installation location (inside the engine, chassis, or body, etc.). Parts with different applications and installation locations are more likely to have surface textures resembling planar tool marks. For example, decorative parts are significantly more likely to have such textures than engine cylinder heads. This functional attribute information can be used to assist in the analysis of in-depth analysis models.

[0044] Furthermore, the deep analysis model in this invention is preferably built based on a general-purpose large-scale model such as DeepSeek or Chart GPT. For example, the aforementioned general-purpose large-scale model can be locally deployed and retrained, and then embedded into the automatic monitoring system of this invention. During use, the feature set of the knife-mark pattern in the affected area, the feature sets of knife-mark patterns in several other areas where planar knife-mark patterns are determined to exist, and the functional attribute information of the automotive part are sent to the locally deployed general-purpose large-scale model in the form of a query. The model's response includes the aforementioned third knife-mark probability. The deep analysis model can also be a vertical model built and trained based on algorithms such as CNN and Transformer; this invention does not specifically limit its application to this.

[0045] Optionally, calculating the second knife-mark probability based on the third knife-mark probability includes: extracting the accessory features of the automotive accessory from any of the high-definition images; determining whether the location belongs to the inner or outer side of the automotive accessory based on the accessory features; configuring a first coefficient if the location belongs to the inner side of the automotive accessory; configuring a second coefficient if the location belongs to the outer side of the automotive accessory; wherein the first coefficient is greater than the second coefficient; and optimizing the third knife-mark probability using the first coefficient or the second coefficient to obtain the second knife-mark probability.

[0046] In this embodiment, certain textures intentionally designed into automotive parts are generally located on the exterior of the parts, and are typically not designed on the interior. For example, decorative textures are usually designed on the outer side of trim panels.

[0047] Based on the aforementioned practical design principles, this invention further extracts the component features of the automotive part from any of the aforementioned high-definition images, including the shape of the component and the position of connectors (e.g., bolt connecting posts / holes). Based on these features, it is possible to determine whether the area containing the suspicious tool marks belongs to the inner or outer side of the automotive part. For example, for an engine cylinder head with a complex shape, by analyzing its image, it can be clearly determined that the side with the bolt connecting posts / holes is the inner side, and the other side is the outer side, thereby determining whether the area containing the suspicious tool marks is the inner or outer side.

[0048] Based on the foregoing analysis, since the probability of the inner and outer sides of automotive parts being misidentified as planar tool marks differs depending on whether they are intentionally designed, this invention further configures different optimization coefficients for the inner and outer sides of automotive parts. Specifically, when the area containing the suspected tool mark is on the inner side of the automotive part, a larger first coefficient is configured, for example, 1.2, which improves the sensitivity of identifying suspected tool marks on the inner side; when the area containing the suspected tool mark is on the outer side of the automotive part, a slightly smaller second coefficient is configured, for example, 1.0 or 1.1, which reduces the sensitivity of identifying suspected tool marks on the outer side.

[0049] Furthermore, since the impact of planar tool marks on the inner side of automotive parts is significantly greater than that on the outer side, a larger first coefficient is assigned to the inner side. This reasonably amplifies the probability of suspicious tool marks on the inner side during calculation, aligning with the reality that the inner side is more sensitive to tool marks and that the damage caused by them is greater. Conversely, a slightly smaller second coefficient is assigned to the outer side, appropriately reducing the weight of the probability of tool marks on the outer side, consistent with the relatively higher tolerance for tool marks on the outer side. It should be noted that the above optimization can be a multiplication operation of the first or second coefficient with the third tool mark probability.

[0050] Optionally, determining the target surface area of ​​the automotive parts based on the location includes: determining the probability difference between the first blade pattern probability and the first probability in the feature set of the blade pattern in the region; matching the surface radius according to the magnitude of the probability difference; wherein the surface radius is positively correlated with the probability difference; and determining the target surface area on the surface of the automotive parts based on the location and the surface radius.

[0051] In this embodiment, as mentioned above, the target surface area should be set to an appropriate size to determine a suitable number of other area tool mark feature sets as auxiliary area tool mark feature sets. Specifically, the present invention first calculates the probability difference between the first tool mark probability and the first probability of the suspected tool mark area. This probability difference is used to represent the degree of closeness between the first tool mark probability and the first probability. The higher the degree of closeness, the more likely the suspected tool mark area is to resemble a planar tool mark. In this case, the surface radius is set to be smaller, that is, only the tool mark feature sets of other areas within a small range that are identified as having planar tool marks are determined as auxiliary area tool mark feature sets. This can reduce the analysis load of the depth analysis model. Conversely, the lower the degree of closeness, the more likely the suspected tool mark area is to resemble an actively designed texture. In this case, more auxiliary area tool mark feature sets need to be called so that the depth analysis model can use more reference data for more accurate and detailed analysis. Through the above adjustment of the surface radius, the present invention achieves a balance between analysis efficiency and analysis accuracy.

[0052] It should be noted that the surface radius mentioned above refers to the actual path along the contour of the outer surface of the automotive part, rather than a straight path in a two-dimensional plane.

[0053] like Figure 3As shown in the figure, this invention also discloses an automatic monitoring system for planar tool marks on automotive parts. The system includes a processing device and a storage device. The computer code stored in the storage device is called and executed by the processing device to achieve the following steps: under uniform light source illumination, high-definition images of automotive parts from multiple angles are captured; tool mark features are extracted from each of the high-definition images to obtain multiple regional tool mark feature sets; wherein, each regional tool mark feature set contains multiple tool mark feature groups, and each tool mark feature group corresponds to a high-definition image at a certain angle; each tool mark feature group in the regional tool mark feature set is input into a classifier, and the multiple tool mark probabilities output by the classifier are processed... The system integrates the features to obtain the first blade pattern probability corresponding to the blade pattern feature set of the region. If the first blade pattern probability is higher than the first probability, it is determined that the automotive part region corresponding to the blade pattern feature set of the region has planar blade patterns. If the first blade pattern probability is lower than the first probability but higher than the second probability, the blade pattern feature set of the region, several other region blade pattern feature sets that are determined to have planar blade patterns, and the functional attribute information of the automotive part are input into the depth analysis model. If the second blade pattern probability output by the depth analysis model is higher than the first probability, it is determined that the automotive part region corresponding to the blade pattern feature set of the region has planar blade patterns. The automotive part regions determined to have planar blade patterns are then labeled.

[0054] This invention also discloses an electronic device, comprising: at least one processor, a memory, and a computer program stored in the memory and executable on the at least one processor, wherein the processor executes the computer program to implement the method described in the foregoing embodiments.

[0055] This invention also discloses a computer storage medium storing a computer program that is executed by a processor to implement the methods described in the foregoing embodiments.

[0056] This invention also discloses a computer program product containing computer code, which, when executed by a processor of an electronic device, implements the method described in the foregoing embodiments.

[0057] The aforementioned computer-readable storage media may include, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, devices, or equipment, or any suitable combination thereof. Alternatively, the computer-readable storage medium may be a machine-readable signal medium. More specific examples of machine-readable storage media include electrical connections based on one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fibers, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination thereof.

[0058] To provide interaction with a user, the systems and techniques described herein can be implemented on an electronic device having: a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user; and a keyboard and pointing device (e.g., a mouse or trackball) through which the user provides input to the electronic device. Other types of devices can also be used to provide interaction with the user; for example, feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including sound input, voice input, or tactile input).

[0059] It should be understood that the various forms of processes shown above can be used, with steps reordered, added, or deleted. For example, the steps described in this invention can be executed in parallel, sequentially, or in different orders, as long as the desired result of the technical solution of this invention can be achieved, and this is not limited herein.

[0060] The specific embodiments described above do not constitute a limitation on the scope of protection of this invention. Those skilled in the art should understand that various modifications, combinations, sub-combinations, and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this invention should be included within the scope of protection of this invention.

Claims

1. An automatic monitoring method for planar tool marks on automotive parts, characterized in that: The method includes the following steps: under uniform light source illumination, high-definition images of automotive parts from multiple angles are captured; knife-mark features are extracted from each high-definition image to obtain multiple regional knife-mark feature sets; wherein, each regional knife-mark feature set contains multiple knife-mark feature groups, and each knife-mark feature group corresponds to a high-definition image at one angle; each knife-mark feature group in the regional knife-mark feature set is input into a classifier, and the multiple knife-mark probabilities output by the classifier are integrated to obtain a first knife-mark probability corresponding to the regional knife-mark feature set; if the first knife-mark probability is higher than the first probability, it is determined that the automotive part region corresponding to the regional knife-mark feature set has planar knife-marks; if the first knife-mark probability is lower than the first probability but higher than the second probability, the regional knife-mark feature set, several other regional knife-mark feature sets that are determined to have planar knife-marks, and the functional attribute information of the automotive part are input into a depth analysis model; if the second knife-mark probability output by the depth analysis model is higher than the first probability, it is determined that the automotive part region corresponding to the regional knife-mark feature set has planar knife-marks; and the automotive part regions determined to have planar knife-marks are labeled. The step of inputting the tool mark feature set of the region, several other region tool mark feature sets that are determined to have planar tool marks, and the functional attribute information of the automotive part into the depth analysis model includes: determining the location of the tool mark feature set of the region on the automotive part, and determining the target surface area of ​​the automotive part based on the location; determining the tool mark feature sets of other regions that are determined to have planar tool marks in the target surface area as auxiliary region tool mark feature sets; inputting the tool mark feature set of the region, each of the auxiliary region tool mark feature sets, and the functional attribute information of the automotive part into the depth analysis model, and the depth analysis model outputting a third tool mark probability; wherein, the functional attribute information includes the application type and installation location of the automotive part; and calculating the second tool mark probability based on the third tool mark probability. The step of calculating the second knife-mark probability based on the third knife-mark probability includes: extracting the accessory features of the car accessory from any of the high-definition images; determining whether the location belongs to the inner or outer side of the car accessory based on the accessory features; configuring a first coefficient if the location belongs to the inner side of the car accessory; configuring a second coefficient if the location belongs to the outer side of the car accessory; wherein the first coefficient is greater than the second coefficient; and optimizing the third knife-mark probability using the first coefficient or the second coefficient to obtain the second knife-mark probability.

2. The automatic monitoring method for planar tool marks on automotive parts according to claim 1, characterized in that: The step of extracting knife-mark features from each of the high-definition images to obtain multiple regional knife-mark feature sets includes: image preprocessing: converting each of the acquired high-definition images to grayscale; removing image noise using Gaussian filtering or median filtering; enhancing image contrast through histogram equalization to highlight texture features; texture region segmentation: extracting texture contours using edge detection algorithms, connecting broken edges using morphological operations to form complete texture regions, and separating texture regions from the background using region growing algorithms or threshold segmentation methods; texture feature extraction: extracting the geometric features of the texture, including direction, spacing, depth, and shape; calculating the texture features of the texture; analyzing the frequency features of the texture using Fourier transform or wavelet transform; integrating the geometric features, texture features, and frequency features of the texture in a single region extracted from each of the high-definition images into the knife-mark feature group, and integrating each of the knife-mark feature groups into a regional knife-mark feature set corresponding to that single region.

3. The automatic monitoring method for planar tool marks on automotive parts according to claim 1, characterized in that: The step of determining the target surface area of ​​the automotive parts based on the location includes: determining the probability difference between the first blade pattern probability and the first probability in the feature set of the blade pattern in the region; matching the surface radius according to the magnitude of the probability difference; wherein the surface radius is positively correlated with the probability difference; and determining the target surface area on the surface of the automotive parts based on the location and the surface radius.

4. An automatic monitoring system for planar tool marks on automotive parts, the system comprising a processing device and a storage device, characterized in that: The computer code stored in the storage device is called and executed by the processing device to achieve the following steps: under uniform light source illumination, high-definition images of automotive parts from multiple angles are captured; knife-mark features are extracted from each of the high-definition images to obtain multiple regional knife-mark feature sets; wherein, each regional knife-mark feature set contains multiple knife-mark feature groups, and each knife-mark feature group corresponds to a high-definition image at one angle; each knife-mark feature group in the regional knife-mark feature set is input into a classifier, and the multiple knife-mark probabilities output by the classifier are integrated to obtain a first knife-mark probability corresponding to the regional knife-mark feature set; if the first knife-mark probability is higher than the first probability, it is determined that the automotive part area corresponding to the regional knife-mark feature set has planar knife-marks; if the first knife-mark probability is lower than the first probability but higher than the second probability, the regional knife-mark feature set, several other regional knife-mark feature sets that are determined to have planar knife-marks, and the functional attribute information of the automotive part are input into a depth analysis model; if the second knife-mark probability output by the depth analysis model is higher than the first probability, it is determined that the automotive part area corresponding to the regional knife-mark feature set has planar knife-marks; the automotive part areas determined to have planar knife-marks are labeled. The step of inputting the tool mark feature set of the region, several other region tool mark feature sets that are determined to have planar tool marks, and the functional attribute information of the automotive part into the depth analysis model includes: determining the location of the tool mark feature set of the region on the automotive part, and determining the target surface area of ​​the automotive part based on the location; determining the tool mark feature sets of other regions that are determined to have planar tool marks in the target surface area as auxiliary region tool mark feature sets; inputting the tool mark feature set of the region, each of the auxiliary region tool mark feature sets, and the functional attribute information of the automotive part into the depth analysis model, and the depth analysis model outputting a third tool mark probability; wherein, the functional attribute information includes the application type and installation location of the automotive part; and calculating the second tool mark probability based on the third tool mark probability. The step of calculating the second knife-mark probability based on the third knife-mark probability includes: extracting the accessory features of the car accessory from any of the high-definition images; determining whether the location belongs to the inner or outer side of the car accessory based on the accessory features; configuring a first coefficient if the location belongs to the inner side of the car accessory; configuring a second coefficient if the location belongs to the outer side of the car accessory; wherein the first coefficient is greater than the second coefficient; and optimizing the third knife-mark probability using the first coefficient or the second coefficient to obtain the second knife-mark probability.

5. An electronic device, comprising: At least one processor, a memory, and a computer program stored in the memory and executable on the at least one processor, characterized in that: the processor executes the computer program to implement the method as claimed in any one of claims 1-3.

6. A computer storage medium storing a computer program, characterized in that: The computer program is executed by a processor to implement the method as described in any one of claims 1-3.

7. A computer program product, characterized in that: The computer program product includes computer code, which, when executed by a processor of an electronic device, implements the method as described in any one of claims 1-3.

Images