Motor vehicle chassis component detection method, apparatus, system, and storage medium

Abstract

This application relates to a method, apparatus, system, and storage medium for detecting motor vehicle chassis components. The method includes: acquiring an image of a target chassis component containing the chassis component of a vehicle to be inspected; extracting visual features of the target chassis component from the image; acquiring a pre-constructed fault feature database, which includes standard fault features and corresponding fault feature judgment conditions; matching the visual features with the standard fault features in the fault feature database to determine the standard fault features that match the visual features, and obtaining the corresponding fault feature judgment conditions based on the matching standard fault features; comparing the visual features with the fault feature judgment conditions, and obtaining the fault determination result of the target chassis component based on the comparison result. By constructing an intelligent detection architecture for chassis component visual feature recognition and fault knowledge base judgment, the method improves the efficiency of motor vehicle chassis detection.

Description

Technical Field

[0001] This application relates to the field of motor vehicle chassis testing technology, and in particular to methods, devices, systems and storage media for testing motor vehicle chassis components. Background Technology

[0002] In the motor vehicle safety technical inspection system, chassis inspection is directly related to the essential guarantee of road driving safety. Its focus is on assessing the condition of key chassis components such as the suspension system, steering system, braking system, exhaust pipe, and fuel lines. The technical condition of these chassis components directly affects the vehicle's handling stability and safety performance.

[0003] Currently, although deep learning-based artificial intelligence technology has been initially applied in limited scenarios such as vehicle appearance recognition and VIN reading, it still faces the problem of low detection efficiency in the vehicle chassis component inspection, which is characterized by complex and variable working environments, a wide variety of inspection objects with different structures, and highly confined working spaces.

[0004] There is currently no effective solution to the problem of low testing efficiency for vehicle chassis components in related technologies. Summary of the Invention

[0005] This embodiment provides a method, apparatus, system, and storage medium for detecting motor vehicle chassis components, in order to solve the problem of low detection efficiency of motor vehicle chassis components in related technologies.

[0006] Firstly, this embodiment provides a method for detecting vehicle chassis components, including:

[0007] Acquire images of target components containing the target chassis components of the vehicle to be inspected;

[0008] Visual features of the target chassis component are extracted from the target component image;

[0009] Obtain a pre-constructed fault feature database, which includes standard fault features and corresponding fault feature judgment conditions;

[0010] The visual features are matched with standard fault features in the fault feature database to determine the standard fault features that match the visual features, and the corresponding fault feature judgment conditions are obtained based on the matching standard fault features.

[0011] The visual features are compared with the fault feature judgment conditions, and the fault judgment result of the target chassis component is obtained based on the comparison result.

[0012] In some embodiments, the fault feature database construction method includes:

[0013] Extract standard fault features from standard fault images of target chassis components;

[0014] Transform fault knowledge in the field of vehicle chassis inspection into structured update rules;

[0015] Based on the structured update rules, generate fault feature judgment conditions corresponding to the standard fault features;

[0016] By associating the standard fault features with the corresponding fault feature judgment conditions, the fault feature database is obtained.

[0017] In some embodiments, extracting the visual features of the target chassis component from the target component image includes:

[0018] If the vehicle under inspection has historical component images containing the target chassis component, compare the historical component images and the target component images to obtain the difference areas between the historical component images and the target component images;

[0019] Based on the difference regions in the target component image, the visual features of the target chassis component are extracted.

[0020] In some embodiments, acquiring a target component image containing a target chassis component of the vehicle to be inspected includes:

[0021] Obtain the chassis image of the vehicle to be inspected;

[0022] The chassis image is divided according to the target chassis component contained in the chassis image to obtain the target component image.

[0023] In some embodiments, acquiring the chassis image of the vehicle to be inspected includes:

[0024] When the vehicle to be inspected enters the preset chassis image acquisition station, chassis images from different angles are obtained.

[0025] In some embodiments, after obtaining chassis images from different angles, the method further includes:

[0026] Preprocessing operations, including perspective correction, illumination equalization, and stitching together chassis images from different angles, yield a preprocessed chassis image.

[0027] In some embodiments, after comparing the visual features with the fault feature judgment conditions and obtaining the fault judgment result of the target chassis component based on the comparison result, the method further includes:

[0028] By integrating the fault determination results of multiple target chassis components contained in the chassis of the vehicle under inspection, a structured inspection report of the chassis of the vehicle under inspection is obtained.

[0029] Secondly, this embodiment provides a vehicle chassis component detection device including: an image acquisition module, a feature extraction module, and a fault determination module; wherein:

[0030] The image acquisition module is used to acquire images of target components, including target chassis components of the vehicle to be inspected.

[0031] The feature extraction module is used to extract visual features of the target chassis component based on the target component image;

[0032] The fault determination module is used to acquire a pre-constructed fault feature database, which includes standard fault features and corresponding fault feature judgment conditions; match the visual features with the standard fault features in the fault feature database to determine the standard fault features that match the visual features, and obtain the corresponding fault feature judgment conditions based on the matching standard fault features; compare the visual features with the fault feature judgment conditions, and obtain the fault determination result of the target chassis component based on the comparison result.

[0033] Thirdly, this embodiment provides a vehicle chassis component testing system, including:

[0034] An image acquisition device is used to acquire images of the target components of the target chassis of the vehicle to be inspected.

[0035] A component inspection device for implementing the steps of the motor vehicle chassis component inspection method described in any one of the first aspects above.

[0036] Fourthly, this embodiment provides a storage medium storing a computer program that, when executed by a processor, implements the steps of the vehicle chassis component detection method described in the first aspect above.

[0037] Compared with related technologies, the vehicle chassis component detection method, apparatus, system, and storage medium provided in this embodiment first acquires an image of the target chassis component containing the target chassis component of the vehicle to be inspected; second, visual features of the target chassis component are extracted from the target component image; next, a pre-constructed fault feature database is acquired, which includes standard fault features and corresponding fault feature judgment conditions; further, the visual features are matched with the standard fault features in the fault feature database to determine the standard fault features that match the visual features, and the corresponding fault feature judgment conditions are obtained based on the matching standard fault features; finally, the visual features are compared with the fault feature judgment conditions, and the fault judgment result of the target chassis component is obtained based on the comparison result. By constructing an intelligent detection architecture for chassis component visual feature recognition and fault knowledge base judgment, it achieves full-process automation from image acquisition to result output, thereby improving the efficiency of vehicle chassis inspection.

[0038] Details of one or more embodiments of this application are set forth in the following drawings and description to make other features, objects and advantages of this application more readily apparent. Attached Figure Description

[0039] The accompanying drawings, which are included to provide a further understanding of this application and form part of this application, illustrate exemplary embodiments and are used to explain this application, but do not constitute an undue limitation of this application. In the drawings:

[0040] Figure 1 This is a hardware structure block diagram of the terminal of a vehicle chassis component testing method according to an embodiment of this application;

[0041] Figure 2 This is a flowchart of a method for testing motor vehicle chassis components according to an embodiment of this application. Figure 1 ;

[0042] Figure 3 This is a flowchart of a method for testing motor vehicle chassis components according to an embodiment of this application. Figure 2 ;

[0043] Figure 4 This is a schematic diagram of the chassis image acquisition layout according to one embodiment of this application;

[0044] Figure 5 This is a structural block diagram of a vehicle chassis component testing device according to an embodiment of this application. Detailed Implementation

[0045] To better understand the purpose, technical solution, and advantages of this application, the application is described and illustrated below in conjunction with the accompanying drawings and embodiments.

[0046] Unless otherwise defined, the technical or scientific terms used in this application shall have the general meaning understood by one of ordinary skill in the art to which this application pertains. Words such as “a,” “an,” “an,” “the,” “the,” and “these” used in this application do not indicate quantitative limitation and may be singular or plural. The terms “comprising,” “including,” “having,” and any variations thereof used in this application are intended to cover non-exclusive inclusion; for example, a process, method, system, product, or device that comprises a series of steps or modules (units) is not limited to the listed steps or modules (units) but may include steps or modules (units) not listed, or may include other steps or modules (units) inherent to these processes, methods, products, or devices. Words such as “connected,” “linked,” and “coupled” used in this application are not limited to physical or mechanical connections but may include electrical connections, whether direct or indirect. “Multiple” used in this application refers to two or more. “And / or” describes the relationship between related objects, indicating that three relationships may exist; for example, “A and / or B” can represent: A alone, A and B simultaneously, and B alone. Normally, the character " / " indicates that the objects before and after it are in an "or" relationship. The terms "first," "second," "third," etc., used in this application are merely to distinguish similar objects and do not represent a specific order of objects.

[0047] The method embodiments provided in this example can be executed on a terminal, computer, or similar electronic device with a certain computing power. For example, it can run on a terminal. Figure 1 This is a hardware structure block diagram of a terminal for a vehicle chassis component testing method according to an embodiment of this application. For example... Figure 1 As shown, a terminal may include one or more ( Figure 1 Only one is shown in the diagram. A processor 102 and a memory 104 for storing data are also included. The processor 102 may be, but is not limited to, a microprocessor (MCU) or a programmable logic device (FPGA). The terminal may also include a transmission device 106 for communication functions and an input / output device 108. Those skilled in the art will understand that… Figure 1 The structure shown is for illustrative purposes only and does not limit the structure of the terminal described above. For example, the terminal may also include components that are larger than... Figure 1 The more or fewer components shown, or having the same Figure 1 The different configurations shown are illustrated.

[0048] The memory 104 can be used to store computer programs, such as application software programs and modules, like the computer program corresponding to the vehicle chassis component detection method in this embodiment. The processor 102 executes various functional applications and data processing by running the computer program stored in the memory 104, thereby implementing the above-described method. The memory 104 may include high-speed random access memory, and may also include non-volatile memory, such as one or more magnetic storage devices, flash memory, or other non-volatile solid-state memory. In some instances, the memory 104 may further include memory remotely located relative to the processor 102, and these remote memories can be connected to the terminal via a network. Examples of such networks include, but are not limited to, the Internet, corporate intranets, local area networks, mobile communication networks, and combinations thereof.

[0049] The transmission device 106 is used to receive or send data via a network. This network includes a wireless network provided by the terminal's communication provider. In one example, the transmission device 106 includes a Network Interface Controller (NIC), which can connect to other network devices via a base station to communicate with the Internet. In another example, the transmission device 106 can be a Radio Frequency (RF) module used for wireless communication with the Internet.

[0050] This embodiment provides a method for testing vehicle chassis components. Figure 2 This is a flowchart of a method for testing motor vehicle chassis components according to an embodiment of this application. Figure 1 ,like Figure 2 As shown, the process includes the following steps:

[0051] Step S210: Obtain an image of the target component containing the target chassis component of the vehicle to be inspected.

[0052] The chassis image data of the vehicle to be inspected is acquired through image acquisition equipment set up at the inspection station. During the acquisition process, the target chassis image can be triggered automatically by the vehicle's arrival sensor, manually by the inspector, or periodically according to a preset inspection procedure. The acquired target chassis image data includes one or more of the following: visible light images, infrared images, or multispectral images, to adapt to different inspection environments and requirements.

[0053] Furthermore, it can perform necessary preprocessing operations on the acquired raw image data, including but not limited to quality optimization processes such as image denoising, contrast enhancement, and illumination equalization, as well as spatial transformation processes such as perspective correction and geometric distortion correction, to obtain clear and standardized images of the target components. Moreover, for detection scenarios requiring coverage of a large chassis area, it can acquire target chassis images from different angles and synthesize a complete chassis image using image stitching technology.

[0054] Based on the acquired chassis images, a pre-trained target detection model is used to automatically identify and locate target chassis components in the chassis images. The target component image regions containing these components are then marked within the chassis images and used as the objects for subsequent analysis and processing. These target chassis components include, but are not limited to, key components of the vehicle chassis such as the left and right front suspensions, left and right rear suspensions, steering tie rods, brake lines, exhaust pipes, fuel lines, stabilizer bars, and engine underbody protection plates.

[0055] It should be noted that the target component image can be a component image obtained directly by taking a picture, an independent image file cropped from the chassis image, or an image region containing the target chassis component that is identified by the target detection model in the chassis image.

[0056] Step S220: Extract visual features of the target chassis component based on the target component image.

[0057] Visual features characterizing the state of the target chassis component are extracted from the target component image. These visual features include one or more of the following state categories: rust, fracture, deformation, breakage, impact, oil leakage, oil stains, and flaking. Visual feature extraction can be achieved using traditional image processing algorithms, machine learning methods, or deep learning methods.

[0058] In one embodiment, the extraction of the above visual features can be performed using corresponding image processing algorithms for different types of visual features: For corrosion features, corrosion areas can be identified by color space conversion combined with threshold segmentation, and the degree of corrosion can be calculated by texture analysis; for fracture features, crack edges can be extracted by edge detection algorithms, linear fracture lines can be detected by Hough transform, and the fracture center line can be obtained based on morphological skeleton extraction; for deformation features, deformation areas can be identified by contour extraction and template matching, and the amount of deformation can be calculated using geometric measurement; for damage features, damage areas can be extracted by image segmentation, and the degree of damage can be judged by edge sharpness analysis; for impact features, dented areas can be identified by height map analysis, and scratch lines can be extracted by directional filtering; for oil leakage features, oil stain areas can be identified by color analysis combined with reflective properties; for oil stain features, the viscosity characteristics of oil stains can be identified by texture analysis, and the thickness of oil stains can be evaluated by multispectral analysis; for detachment features, missing areas can be identified by image difference and template comparison, and the integrity of the attachment can be judged by contextual reasoning.

[0059] Furthermore, a unified object detection algorithm can be used to simultaneously identify multiple visual features. Based on deep learning methods, the object detection model pre-trained can treat various visual features as different detection categories, enabling the model to simultaneously locate and identify multiple visual features in an image, such as rusted areas, cracks, deformed parts, damaged locations, dents, oil leaks, oil stains, and missing parts, in a single forward computation. The object detection model can be implemented using algorithms such as YouOnly Look Once (YOLO), Single Shot MultiBox Detector (SSD), and RetinaNet. Through end-to-end learning, the model directly outputs the location and category information of various visual features from the image.

[0060] As another way to extract visual features, a multi-task learning framework can be used to integrate multiple tasks such as feature region detection, feature classification, and feature quantization into the same model. Multiple task branches are set on the basis of a shared feature extraction network, and the location and category information of various visual features are output at the same time.

[0061] In addition, a hybrid recognition strategy can be formed by combining traditional image processing algorithms with deep learning models. This involves first using a deep learning model for initial feature region detection, followed by refined analysis and quantification of the detected regions using specialized image processing algorithms; or first using traditional algorithms to extract candidate feature regions, followed by confirmation and classification of these regions using a deep learning model. Ultimately, accurate location and category information for various visual features are obtained.

[0062] The aforementioned visual features can be comprehensively represented and stored in multiple forms, including visualizing the distribution and type of visual features on components in the form of heat maps and annotation maps, combining various visual features into multi-dimensional feature vectors as input for subsequent matching analysis, and recording the attributes of each visual feature in the form of structured data, including feature type, location coordinates, and confidence level.

[0063] Step S230: Obtain a pre-built fault feature database, which includes standard fault features and corresponding fault feature judgment conditions.

[0064] A pre-built fault feature database is acquired. This database is a collection of data used to store and manage fault feature information and fault feature judgment conditions for various target chassis components. The fault feature database records faults according to component type and fault type. Each fault record contains at least two core elements: first, a standard fault feature, which characterizes the component type and fault type; and second, the fault feature judgment conditions corresponding to the standard fault feature, which are the quantitative basis or judgment rules used to determine whether the current component has reached the fault state.

[0065] Specifically, standard fault features are visual representations extracted from chassis component sample images that have been labeled with standard fault features. Taking common fault types in chassis inspection as examples, standard fault features include, but are not limited to: "Shock absorber oil leakage" is characterized by visual features such as oil stains, oil flow marks, or oil droplets adhering to the surface of the shock absorber cylinder; "Exhaust pipe rust perforation" is characterized by visual features such as flaky rust areas, rust perforations, and irregularly shaped holes on the exhaust pipe wall; "Stabilizer bar linkage looseness" is characterized by visual features such as abnormal clearance between the linkage and the connection, linkage angle deviation, bushing wear, or displacement. In addition to the above types, standard fault features also include linear fracture features corresponding to "brake hose cracks," bending or twisting features corresponding to "steering tie rod deformation," separation features at the connection point corresponding to "fuel hose detachment," and dents or damage features corresponding to "engine underbody protection plate impact," among other standard visual representations of various fault types.

[0066] Fault feature judgment conditions are the criteria associated with various standard fault features, used to compare the currently extracted visual features with the standard fault features to make a status determination. The forms of fault feature judgment conditions include: First, quantitative threshold-based judgment conditions, such as setting thresholds for the proportion of rust area to the total area of ​​the component, rust depth, or rust area color depth for rust-related faults; setting thresholds for deformation amount, bending angle, or straightness deviation for deformation-related faults; and setting thresholds for oil stain area, oil stain coverage density, or leakage trace length for oil leakage-related faults. Second, similarity threshold-based judgment conditions, which calculate the similarity between the current visual features and standard fault features. When the similarity exceeds a preset threshold, it is determined to match the fault type. Similarity calculation can use Euclidean distance, cosine distance, Mahalanobis distance, or distance metrics based on deep learning features. Third, rule-based judgment conditions, which use a combination of fault feature judgment rules to determine faults. For example, "If an oil stain area exists simultaneously, and the oil stain area is located in the middle of the shock absorber cylinder and the oil stain area exceeds a set value, then it is determined that the shock absorber is leaking oil." Rules can take the form of logical expressions, decision tree rules, or fuzzy logic rules. Fourth, classification model judgment conditions, which use the trained classifier parameters as judgment conditions. The current visual features are input into the classifier model, and the fault type and severity are determined based on the model's output. Classifiers include support vector machines, random forests, Naive Bayes, or neural networks.

[0067] The construction of a fault feature database can be achieved through various methods, including machine learning and expert system construction. First, machine learning methods are used to train a large number of chassis component images labeled with fault types. Standard feature representations for various faults are automatically learned from the training samples, and corresponding fault feature judgment conditions are determined. Second, an expert system is used to receive fault knowledge and judgment experience input from domain experts. This fault knowledge and judgment experience are transformed into structured standard fault features and fault feature judgment conditions, which are then stored in the database after validity verification. Validity verification refers to the process of performing format standardization checks, logical consistency checks, accuracy verification based on test samples, expert cross-validation, and retrospective verification with historical data on newly input fault knowledge or update rules. This ensures the accuracy and consistency of fault knowledge or update rules and avoids erroneous or invalid fault features entering the database. Finally, the above machine learning and expert system construction methods can be combined. Initial fault features are first learned from samples using machine learning, and then the standard fault features and fault feature judgment conditions are optimized, adjusted, and continuously updated using an expert system.

[0068] The fault feature database supports dynamic updates and maintenance. When new fault types emerge or existing fault feature descriptions need optimization, new standard fault features and their corresponding fault feature judgment conditions can be added to the database through incremental learning, or existing standard fault features and fault feature judgment conditions can be corrected and adjusted. An update log is recorded during the database update process, including update time, update content, update source, and other information, supporting version rollback and change tracking.

[0069] Step S240: Match the visual features with the standard fault features in the fault feature database to determine the standard fault features that match the visual features, and obtain the corresponding fault feature judgment conditions based on the matching standard fault features.

[0070] The visual features of the target chassis component extracted in step S220 are matched with the standard fault features in the fault feature database obtained in step S230. Based on the matching results, the standard fault features that match the current visual features are determined. Then, the fault feature judgment conditions corresponding to the standard fault features are obtained from the fault feature database as the basis for subsequent judgment.

[0071] When performing feature matching, an appropriate matching strategy is selected based on the type and representation of the visual features. These matching strategies include: global matching, hierarchical matching, multi-scale matching, attention-guided matching, and multimodal matching. When determining the similarity calculation method, an appropriate similarity calculation method is selected based on the type of visual features and the matching strategy. These methods include distance-based similarity calculation, correlation-based similarity calculation, information theory-based similarity calculation, kernel function-based similarity calculation, probability-based similarity calculation, and deep learning-based similarity calculation. When selecting standard fault features for matching based on the similarity calculation results, corresponding matching conditions are set, including threshold conditions, optimal conditions, top k selection conditions, clustering conditions, multi-level screening conditions, and confidence conditions. After matching is completed, the matching results can be verified and optimized through methods such as cross-validation, temporal consistency verification, spatial consistency verification, and human-machine collaborative verification.

[0072] Based on the established standard fault characteristics, corresponding fault characteristic judgment conditions are retrieved from the fault characteristic database. These judgment conditions can be a single condition, a combination of multiple conditions, or a set of conditions, including one or more of the following: quantification threshold-based judgment conditions, similarity threshold-based judgment conditions, rule-based judgment conditions, and classification model judgment conditions. These serve as the direct basis for further fault determination. When the fault characteristic judgment conditions are a combination of multiple conditions and the judgment results of each condition are inconsistent, multiple judgment results can be comprehensively processed according to a preset fusion strategy. This includes adopting the majority consensus conclusion, simultaneously outputting multiple fault types and their confidence levels, or pushing contradictory results to a manual review terminal for final confirmation by testing personnel.

[0073] Step S250: Compare the visual features with the fault feature judgment conditions, and obtain the fault judgment result of the target chassis component based on the comparison result.

[0074] Based on the type of fault feature judgment conditions obtained in step S240, the visual features extracted in step S220 are compared with the fault feature judgment conditions using the corresponding comparison strategy. The fault judgment result of the target chassis component is determined based on the comparison result, and the corresponding fault information is output.

[0075] The acquired standard fault features are identified, and different judgment conditions are used for different standard fault features. Fault feature judgment conditions can include quantification threshold judgment conditions, similarity threshold judgment conditions, rule-based judgment conditions, classification model judgment conditions, and reference judgment conditions, etc.

[0076] Specifically, for quantification threshold-based judgment conditions, the corresponding quantification indicators in the visual features are numerically compared with preset thresholds. Quantification indicators can include percentage of rust area, crack length, deformation, oil stain coverage area, etc. The comparison result directly determines whether a fault state has been reached; for multi-level thresholds, the severity level of the fault can also be determined. For similarity threshold-based judgment conditions, the similarity value between the calculated visual features and standard fault features is compared with a preset similarity threshold. For rule-based judgment conditions, the visual features are converted into factual assertions and input into a rule engine for rule matching. The rule engine infers based on preset fault feature judgment conditions; when the rule conditions are met, the corresponding rule is activated, and the fault judgment result is obtained based on the conclusion of the activated rule. When multiple rules are activated simultaneously, conflict resolution strategies such as priority ranking and rule specificity are used to determine the final rule adopted. For classification model judgment conditions, the visual features are input into a pre-trained classification model for calculation, and the fault type and fault confidence are determined based on the model's output classification result. For reference-based judgment conditions, the visual features in the target component image area are compared with reference objects, including standard reference images of the same model in normal condition and historical inspection images of the vehicle. The changing trend of the component's condition is identified by image difference comparison or feature difference calculation, such as the expansion of rust area or the appearance of new cracks.

[0077] Due to the limited working space of vehicle chassis, inspectors must conduct visual inspections from pits or under lifts, resulting in high labor intensity, poor working conditions, and difficulty in accurately assessing the condition of components in concealed areas and confined spaces. Inspection efficiency is highly dependent on the experience level of the inspectors, leading to strong subjectivity, poor consistency, and a high risk of missed or false positives. Traditional inspection methods based on fixed templates or rules are ill-suited to the diverse inspection needs of vehicles. While deep learning-based artificial intelligence technology has seen initial applications in limited scenarios such as vehicle appearance recognition and VIN reading, it still faces challenges in inspecting vehicle chassis components in harsh working environments with uneven lighting, oil and mud adhesion, and rust reflection interference, resulting in low inspection efficiency.

[0078] Steps S210 to S250 above, by constructing an intelligent detection architecture for visual feature recognition of target chassis components and fault knowledge base determination, realize full-process automation from image acquisition to result output, thereby improving the efficiency of motor vehicle chassis inspection.

[0079] In one embodiment, the method for constructing a fault feature database includes:

[0080] (1) Extract standard fault features from standard fault images of target chassis components.

[0081] The standard fault images can be typical sample images selected from a large number of real-world chassis component fault cases, simulated fault state images generated through simulation modeling, or example images obtained from technical standards and maintenance manuals. For each standard fault image, its corresponding standard fault characteristics are labeled. These standard fault characteristics include, but are not limited to, "shock absorber oil leakage," "exhaust pipe rust penetration," and "stabilizer bar connecting rod looseness." For the same standard fault characteristic, fault features can be extracted from multiple standard fault images at different angles, with varying degrees of severity, and under different lighting conditions to form a standard fault feature set for that standard fault characteristic.

[0082] (2) Transform fault knowledge in the field of motor vehicle chassis inspection into structured update rules.

[0083] The fault knowledge comes from the experience of domain experts, technical standards and specifications, maintenance manuals, fault case databases, etc., including the definition of fault types in the field of motor vehicle chassis inspection, typical visual manifestations of faults, fault judgment criteria, fault severity classification, and the correspondence between faults and component types.

[0084] Structured update rules refer to the transformation of fault knowledge into a standardized expression that can be recognized and processed by computers. Structured update rules can be expressed in various forms, including condition-conclusion, logical expression, decision tree, production rule, or framework representation. Each structured update rule includes attribute information such as rule identifier, target chassis component type, fault type, fault feature judgment conditions, fault feature judgment threshold, rule priority, rule confidence, and rule validity period.

[0085] Optionally, the aforementioned fault knowledge can be transformed into structured update rules in a standardized expression form that can be recognized and processed by computers by defining standardized knowledge templates, using structured description languages ​​for field encoding, and establishing terminology mapping tables. For example, expert experience describing "significant oil leakage in the left front shock absorber" can be converted into a structured update rule with the following structure: "Rule ID: RULE_001; Target component: Left front shock absorber; Fault type: Shock absorber oil leakage; Rule conditions: Existence of an oil stain area located in the middle or lower part of the shock absorber cylinder; Judgment threshold: Oil stain area greater than 5% of the component area or oil stain length greater than 3cm; Rule priority: High; Confidence: 0.95".

[0086] (3) Generate fault feature judgment conditions corresponding to standard fault features according to the structured update rules.

[0087] Optionally, the standardized expression in the structured update rules is parsed, and the qualitative descriptions in the structured update rules are converted into quantifiable judgment indicators to generate specific fault feature judgment conditions corresponding to the standard fault features. For example, based on the above structured update rules, the fault feature judgment conditions corresponding to the standard fault feature of shock absorber oil leakage are generated, including "oil stain area threshold: the percentage of the oil stain area to the visible area of ​​the shock absorber cylinder is greater than 5%" and "oil stain length threshold: the flow length of the oil stain along the cylinder direction is greater than 3cm".

[0088] (4) Associate the standard fault features and the corresponding fault feature judgment conditions to obtain the fault feature database.

[0089] The extracted standard fault features are mapped to the generated fault feature judgment conditions to form a complete fault feature record. All fault feature records are saved according to dimensions such as target chassis component, fault type, and vehicle type, and a multi-dimensional indexing mechanism is established to form a fault feature database that can be used for subsequent matching and judgment. After the database is built, the fault features and judgment conditions in the database can be verified and optimized using a test sample set to ensure their accuracy and effectiveness.

[0090] By using structured update rules derived from fault knowledge, standard fault features are extracted and associated with fault feature judgment conditions. This establishes the dependency relationship between fault features and judgment conditions, eliminating the need to reconstruct the entire set of judgment conditions when updating fault samples, thus reducing the maintenance cost and update difficulty of the fault feature database.

[0091] In addition, in one embodiment, extracting visual features of the target chassis component based on the target component image includes: if there is a historical component image of the vehicle under inspection containing the target chassis component, comparing the historical component image and the target component image to obtain the difference region between the historical component image and the target component image; and extracting visual features of the target chassis component based on the difference region in the target component image.

[0092] If the vehicle to be inspected has historical inspection records containing the target chassis component, first retrieve the chassis images collected and stored during each inspection or repair of the vehicle from the historical database, and locate the historical component image that is the same as the current target chassis component based on the historical annotation information.

[0093] Secondly, the historical component image is compared with the currently acquired target component image to obtain the difference regions between them. Optionally, a feature-point-based image registration technique is used to calculate the difference regions between the historical component image and the currently acquired target component image through feature point matching. Specifically, in one embodiment, firstly, a geometric transformation is performed on the historical component image or the target component image according to the spatial transformation parameters obtained from the matching calculation, so that the historical component image and the currently acquired target component image are aligned in spatial position. Subsequently, pixel-level or feature-level difference operations are performed on the aligned historical component image and the currently acquired target component image to calculate the grayscale difference, color difference, or texture difference at the corresponding pixel positions. Next, threshold segmentation and morphological processing are performed on the difference results to extract pixel regions with significant differences as candidate difference regions. Finally, connected component analysis and screening are performed on the candidate difference regions to remove noise points and isolated points, merge adjacent regions, and obtain the final difference regions. The difference regions reflect the changes in the component's state over time, including areas of expanded rust area, newly appearing crack areas, areas of intensified deformation, newly added oil stain areas, and displacement areas caused by component loosening, etc.

[0094] Next, based on the difference regions in the target component image, visual features of the target chassis component are extracted. In one embodiment, the identified difference regions are designated as key areas of interest, and local feature extraction of these regions can be performed in the target component image using traditional image processing algorithms, machine learning methods, or deep learning methods. The extracted visual features include the area, perimeter, shape, color, and texture of the difference regions, as well as their relative coordinates on the component. For example, for difference regions reflecting rust expansion, rust color depth and rust morphology features are extracted; for difference regions reflecting newly appearing cracks, crack length, width, and orientation angle are extracted; for difference regions reflecting aggravated deformation, deformation amount and deformation direction are extracted; for difference regions reflecting newly added oil stains, oil stain area, oil stain color depth, and oil stain morphology features are extracted; and for difference regions reflecting component loosening, displacement and angular offset are extracted.

[0095] Finally, for cases where historical component images are unavailable, the overall visual features are extracted directly from the target component image without historical comparison or difference analysis.

[0096] In this embodiment, by comparing historical component images with target component images, the system identifies areas of change in the target chassis component's condition. This effectively captures progressive fault characteristics such as rust expansion, crack propagation, and new oil stains, avoiding missed detections or misjudgments that might occur if only a single image is used. Simultaneously, the identified difference areas are used for local feature extraction, highlighting the weight of fault-affected areas while reducing interference from irrelevant regions in the image, thus improving the accuracy of vehicle chassis fault detection. Furthermore, the comparison results between historical component images and target component images provide data support for fault evolution trend analysis, enabling the creation of a digital chassis file for each vehicle, and facilitating full lifecycle traceability and dynamic monitoring of the target chassis component's condition.

[0097] In one embodiment, obtaining a target component image containing a target chassis component of a vehicle to be inspected includes: obtaining a chassis image of the vehicle to be inspected; and dividing the chassis image according to the target chassis component contained in the chassis image to obtain a target component image.

[0098] After acquiring the chassis image of the vehicle to be inspected, the chassis image is segmented according to the target chassis components contained within it to obtain the target component image. The segmentation method can employ traditional segmentation methods based on image processing techniques; it can also employ object detection methods based on deep learning; alternatively, candidate regions can be quickly extracted using traditional segmentation methods, and then classified and located using a deep learning model; or preliminary detection can be performed using a deep learning model, followed by fine-tuning of the detection boundaries using traditional segmentation methods.

[0099] In one embodiment, a deep learning-based object detection algorithm can be used to automatically identify and locate various components in a chassis image, thereby achieving image segmentation. Specifically, a deep learning object detection model is constructed and trained. The model takes the target component image as input and outputs the positional and category information of each target chassis component in the target component image. During training, a large number of chassis images of different vehicle models, angles, and lighting conditions are collected as training samples. The target chassis components in the sample images are manually labeled. The labeled component categories can include left and right front suspensions, left and right rear suspensions, steering tie rods, brake lines, exhaust pipes, fuel lines, stabilizer bars, and engine underbody protection plates, etc. The labeling information includes the bounding box coordinates of each component in the image and the component category label. The labeled training samples are input into the object detection network for iterative training. The network parameters are continuously optimized through the backpropagation algorithm, enabling the model to learn the appearance and spatial position features of each component.

[0100] After training, the chassis image of the vehicle to be inspected is input into the target detection model for forward inference calculation. The model first extracts multi-scale feature maps of the image through a backbone network, which can adopt a convolutional neural network structure. Then, candidate regions are generated through a region proposal network or anchor point mechanism. Finally, the model performs component category discrimination and bounding box location confirmation for each candidate region through classification and regression branches. The model output includes the bounding box coordinates, category label, and detection confidence score for each detected target chassis component. Reliable detection results are selected based on a preset confidence threshold. By dividing the original chassis image according to the bounding box coordinates of each component, a series of target component images corresponding to each target chassis component can be obtained.

[0101] Object detection algorithms can employ either single-stage or two-stage detectors. Single-stage detectors include YOLO, SSD, and RetinaNet, characterized by their high detection speed, making them suitable for real-time processing. Two-stage detectors include Faster Region-based Convolutional Neural Networks (Faster R-CNN) and Mask Region-based Convolutional Neural Networks (Mask R-CNN), characterized by their high detection accuracy and ability to more accurately locate part boundaries. The specific algorithm chosen depends on the required detection accuracy and processing time.

[0102] By automatically segmenting chassis images using target detection algorithms, individual target component images can be quickly and accurately separated from complex chassis images, laying the foundation for subsequent component-level feature extraction and fault diagnosis. Furthermore, the segmented target component images have simple backgrounds and clearly defined targets, effectively avoiding mutual interference between different components and improving the targeting of feature extraction and the accuracy of fault diagnosis.

[0103] In one embodiment, acquiring chassis images of the vehicle to be inspected includes: acquiring chassis images from different angles when the vehicle to be inspected enters a preset chassis image acquisition station.

[0104] Specifically, once the vehicle reaches the designated position and triggers the positioning sensor, the image acquisition process automatically begins. This process controls an industrial camera array positioned within the lift pit, on the moving track, or at multiple fixed locations to simultaneously or sequentially capture images from multiple preset angles under the vehicle, obtaining a comprehensive image covering the entire chassis. The image acquisition device can be configured differently depending on the workstation layout. For example, for inspection workstations with pits, the camera array is fixedly installed on both sides and the bottom of the pit, capturing images from below; for inspection workstations with a planar layout, a mobile scanning mechanism drives the camera to continuously capture images along the vehicle's underside trajectory; for special inspection needs, handheld or vehicle-mounted imaging devices can be used for supplementary acquisition. During the acquisition process, the camera aperture, shutter speed, and supplementary lighting are adjusted to ensure clear and uniform chassis images are obtained under different lighting conditions.

[0105] By using a multi-angle image acquisition configuration, it can comprehensively cover the visual information of all areas of the chassis, effectively solving the problem of components occluding each other and limiting the field of vision under a single viewpoint, and ensuring that key components such as the suspension system, brake lines, and exhaust pipes can be clearly acquired from different positions and angles.

[0106] In one embodiment, after obtaining chassis images from different angles, the method for detecting vehicle chassis components further includes: obtaining a preprocessed chassis image through preprocessing operations such as perspective correction, illumination equalization, and stitching together chassis images from different angles.

[0107] Optionally, firstly, perspective correction is performed on the acquired original chassis images from various angles. Due to the influence of shooting angle and lens distortion, originally parallel lines in the original chassis images may converge, and the actual shape and size of the components will be distorted by perspective. For the original chassis images, lens distortion correction is performed using pre-calibrated camera intrinsic parameters and distortion coefficients to eliminate radial and tangential distortion. Simultaneously, feature points or reference calibration objects in the original chassis images are extracted, and the perspective transformation matrix is ​​calculated to project the original chassis images from the original viewpoint to a unified vertical or horizontal viewpoint, restoring the true geometric relationships of the components in the original chassis images and ensuring the accuracy of subsequent measurements and analyses.

[0108] Secondly, the perspective-corrected image undergoes illumination equalization processing. The chassis operates in a complex environment with uneven lighting conditions, and different areas may exhibit issues such as shadows, reflections, excessive darkness, or excessive brightness, affecting the consistency of image quality. Subsequently, brightness histogram analysis is performed on the image to identify areas of abnormal lighting. An adaptive histogram equalization algorithm is employed to adjust the contrast of local areas, enhancing details in dark areas while suppressing excessively bright areas. Furthermore, a multi-scale image enhancement algorithm is combined to decompose the image into illumination and reflection components. While maintaining the reflection component unchanged, the illumination component is corrected to eliminate the effects of uneven lighting.

[0109] Finally, the multi-angle chassis images, after perspective correction and illumination equalization, are stitched together to form a complete chassis image. Feature points are extracted from the chassis images, and spatial transformation relationships between adjacent images are calculated using a feature point matching algorithm, including translation, rotation, and scaling parameters. Images are aligned according to these transformation parameters to eliminate positional deviations in overlapping areas. Weighted fusion, multi-band fusion, or Laplacian pyramid fusion algorithms are employed to smoothly transition overlapping areas, eliminating stitching artifacts and ensuring the visual continuity and consistency of the chassis image.

[0110] After preprocessing, a complete, clear, uniformly illuminated, and geometrically accurate preprocessed chassis image is obtained. The preprocessed chassis image eliminates various distortions and quality defects in the original chassis image, organically integrating chassis images taken from all angles into one, providing a clear and accurate data foundation for subsequent component identification, feature extraction, and fault determination, while also facilitating human observation and manual verification.

[0111] In addition, in one embodiment, after comparing visual features with fault feature judgment conditions and obtaining the fault judgment result of the target chassis component based on the comparison result, the vehicle chassis component detection method further includes: integrating the fault judgment results of multiple target chassis components contained in the chassis of the vehicle under inspection to obtain a structured inspection report of the chassis of the vehicle under inspection.

[0112] Specifically, firstly, the fault assessment results of all target chassis components are collected and summarized. The fault assessment results obtained from each target chassis component are then collected uniformly. All components identified during the inspection process, regardless of their status (normal, abnormal, or severely faulty), are included in the summary to form a complete component status list. Secondly, correlation analysis and logical verification can be performed on the fault assessment results of each component. The logical correlation or causal relationship between the fault assessment results of different components is checked. For example, deformation of the steering tie rod may be related to loosening of the stabilizer bar link, and leakage of the brake line may be related to impact with the upper component. For related faults, correlation marking and prompts are made. Simultaneously, the consistency of the assessment results is verified to ensure that the assessment results of the same component in target chassis component images from different angles corroborate each other, avoiding contradictory assessments. Assessment results with low confidence or logical doubts are marked as pending review or require manual confirmation.

[0113] Furthermore, a structured inspection report is generated. This report includes basic vehicle information, inspection results, a detailed list of components, trend analysis, and repair recommendations. The basic vehicle information includes license plate number, vehicle model, and inspection time. The inspection results are presented as a summary of the overall chassis inspection conclusions, including the total number of inspected components, the number of normal components, the number of abnormal components, the number of seriously faulty components, and key safety hazard warnings. The component details list displays detailed assessment results for each component, grouped by chassis region or component category, including component name, fault type, fault severity, and confidence level, with seriously faulty components highlighted. The trend analysis, for vehicles with historical inspection records, shows the changing trends of key component conditions over time, such as rust area change curves, crack propagation, and oil leak evolution. Repair recommendations are automatically generated based on the fault assessment results, including suggested repair items, repair priority, suggested repair timeframe, reference repair solutions, and required spare parts information. The structured inspection report is output in various formats to meet different application needs.

[0114] Finally, the inspection report is stored in the vehicle chassis historical image database, and the "one vehicle, one file" digital chassis file is updated simultaneously. The generated inspection report, along with the collected chassis images, information on each component within the chassis images, and intermediate processing results, are all stored in the vehicle's historical record, forming a complete inspection record chain. The updated "one vehicle, one file" file contains complete data from all previous vehicle inspections, supporting historical review, trend analysis, and data mining, providing comprehensive data support for vehicle annual inspections, maintenance, and insurance claims.

[0115] By integrating and summarizing the fault determination results of multiple target chassis components, a structured inspection report is generated, realizing the organic integration from single component determination to whole vehicle chassis condition assessment, and providing inspection personnel with comprehensive and intuitive inspection conclusions.

[0116] Figure 3 This is a flowchart of a method for testing motor vehicle chassis components according to an embodiment of this application. Figure 2 .like Figure 3 As shown, the method for testing vehicle chassis components includes the following steps:

[0117] Step S301 involves simultaneously capturing images of multiple zones of the vehicle chassis under inspection using multiple high-definition industrial cameras at the inspection station. In one embodiment, six high-definition industrial cameras are used to capture images of six zones.

[0118] In step S302, the image processing server receives the original images from multiple partitions, performs image stitching, correction, and image preprocessing to generate a complete chassis image.

[0119] Step S303: Target detection is performed on the chassis image using a target detection model deployed on the server to obtain target component images. The target detection model is trained on tens of thousands of labeled chassis component images, automatically identifying 15 key chassis components, including the "left front shock absorber," "exhaust pipe midsection," and "brake oil pipe," and marking the specific spatial location of each component with bounding boxes to obtain the target component image corresponding to each target chassis component.

[0120] Step S304: Based on the license plate number of the vehicle to be inspected, retrieve the chassis image of the vehicle from the vehicle chassis historical image database from the last inspection, and automatically locate the historical component image of the same component in the target component image.

[0121] Step S305: Compare the historical component image with the target component image using an image difference algorithm to obtain the difference region between the two images. Taking the "left front shock absorber" as an example, the comparison reveals a new dark oil stain area on the lower part of the shock absorber dust cover in the target component image, and the visual features of this difference region are extracted.

[0122] Step S306: Match the visual features of the discrepancy area with standard fault features in a pre-built fault feature database. The matching result shows that the features of the oil stain area match the standard fault feature of "shock absorber oil leakage" in the database.

[0123] Step S307: Based on the fault feature judgment conditions corresponding to the standard fault features obtained through matching, the visual features of the difference area are compared with the judgment conditions, and the state of the target chassis component is determined based on the comparison results. In this embodiment, the feature state of the oil stain area is "abnormal: shock absorber suspected of leaking oil".

[0124] Step S308: Integrate the fault determination results of multiple target chassis components included in the chassis of the vehicle under inspection to generate a structured inspection report. The report highlights the oil-leaking shock absorber area on the chassis image and provides a prompt: "Compared to the last inspection (October 2023), new oil stains have appeared on the left front shock absorber, and inspection is recommended."

[0125] Step S309: All the original images, processed images, analysis results and test reports collected are stored in the chassis history image database of the vehicle to complete the update of the "one vehicle, one file" digital chassis file.

[0126] Steps S301 to S309 above utilize a fully automated chassis component inspection method to achieve a complete closed loop from image acquisition, preprocessing, component identification, historical comparison, fault determination to report generation and file updating. The entire inspection process is shortened from the original time-consuming manual operation to automated processing in a few minutes. While ensuring inspection accuracy, it improves inspection efficiency and meets the actual needs of efficient chassis inspection in scenarios such as annual vehicle inspection and used car evaluation.

[0127] Figure 4 This is a schematic diagram of the chassis image acquisition layout according to an embodiment of this application, as shown below. Figure 4 As shown, the chassis image acquisition system includes high-definition industrial cameras 41, strip lights 42, and inductive loops 43. Specifically, there are six high-definition industrial cameras 41, providing comprehensive coverage of the entire vehicle body. These cameras utilize high-resolution global shutter technology, effectively suppressing image blur caused by vehicle movement and ensuring the clarity and accuracy of the acquired images. Two strip lights 42, distributed on the left and right sides of the vehicle body, provide uniform and stable supplementary lighting in environments with insufficient or uneven lighting. These strip lights use high color rendering index (CRI) light sources, accurately reproducing the original colors of chassis components and avoiding color distortion caused by lighting issues. Simultaneously, adjusting the brightness of the lights effectively suppresses common shadow areas under the chassis and eliminates interference factors such as oil stain reflections and metal reflections, providing high-quality original images for subsequent image processing. The inductive loops 43 are buried under the ground at the inspection station to detect whether the vehicle has reached the designated shooting position. When the vehicle enters the inspection station and passes over the inductive coil 43, the inductive coil 43 senses the change in inductance caused by the vehicle's metal components, generates a trigger signal and sends it to the control system. The control system then automatically starts the high-definition industrial camera 41 and the strip fill light 42 to acquire images, realizing automatic trigger shooting without human intervention.

[0128] Based on the same inventive concept, this application also provides a vehicle chassis component testing device for implementing the above-described vehicle chassis component testing method. The solution provided by this device is similar to the solution described in the above method; therefore, the specific limitations in one or more embodiments of the vehicle chassis component testing device provided below can be found in the limitations of the vehicle chassis component testing method described above, and will not be repeated here.

[0129] Figure 5 This is a structural block diagram of a motor vehicle chassis component testing device 50 according to an embodiment of this application, as shown below. Figure 5As shown, the vehicle chassis component detection device 50 includes: an image acquisition module 52, a feature extraction module 54, and a fault determination module 56; wherein: the image acquisition module 52 is used to acquire an image of a target component containing the target chassis component of the vehicle to be inspected; the feature extraction module 54 is used to extract visual features of the target chassis component based on the target component image; the fault determination module 56 is used to acquire a pre-built fault feature database, which includes standard fault features and corresponding fault feature judgment conditions; match the visual features with the standard fault features in the fault feature database to determine the standard fault features that match the visual features, and obtain the corresponding fault feature judgment conditions based on the matching standard fault features; compare the visual features with the fault feature judgment conditions, and obtain the fault determination result of the target chassis component based on the comparison result.

[0130] It should be noted that the above modules can be functional modules or program modules, and can be implemented through software or hardware. For modules implemented through hardware, the above modules can reside in the same processor; or the above modules can be located in different processors in any combination.

[0131] This embodiment also provides a vehicle chassis component inspection system, including an image acquisition device and a component inspection device. The image acquisition device is used to acquire target component images of the target chassis component of the vehicle to be inspected; the component inspection device is used to implement the steps in any of the above method embodiments.

[0132] Optionally, the electronic device may further include a transmission device and an input / output device, wherein the transmission device is connected to the image acquisition device and the input / output device is connected to the component detection device.

[0133] Optionally, in this embodiment, the component detection device described above can be configured to perform the following steps via a computer program:

[0134] S1, Obtain the target component image containing the target chassis component of the vehicle to be inspected;

[0135] S2, Extract visual features of the target chassis component from the target component image;

[0136] S3, Obtain a pre-built fault feature database, which includes standard fault features and corresponding fault feature judgment conditions;

[0137] S4. Match the visual features with the standard fault features in the fault feature database to determine the standard fault features that match the visual features, and obtain the corresponding fault feature judgment conditions based on the matching standard fault features.

[0138] S5 compares the visual features with the fault feature judgment conditions, and obtains the fault judgment result of the target chassis component based on the comparison result.

[0139] It should be noted that the specific examples in this embodiment can refer to the examples described in the above embodiments and optional implementations, and will not be repeated in this embodiment.

[0140] Furthermore, in conjunction with the vehicle chassis component detection method provided in the above embodiments, this embodiment can also provide a storage medium for implementation. This storage medium stores a computer program; when executed by a processor, the computer program implements any of the vehicle chassis component detection methods described in the above embodiments.

[0141] It should be understood that the specific embodiments described herein are merely illustrative of the application and not intended to limit it. All other embodiments derived by those skilled in the art based on the embodiments provided in this application without inventive effort are within the scope of protection of this application.

[0142] Obviously, the accompanying drawings are merely some examples or embodiments of this application. Those skilled in the art can apply this application to other similar situations based on these drawings without any creative effort. Furthermore, it is understood that although the work done in this development process may be complex and lengthy, for those skilled in the art, certain design, manufacturing, or production modifications made based on the technical content disclosed in this application are merely conventional technical means and should not be considered as insufficient disclosure of this application.

[0143] The term "embodiment" in this application refers to a specific feature, structure, or characteristic described in connection with an embodiment that may be included in at least one embodiment of this application. The appearance of this phrase in various places in the specification does not necessarily imply the same embodiment, nor does it imply that it is mutually exclusive with or independent of other embodiments. It will be clearly or implicitly understood by those skilled in the art that the embodiments described in this application may be combined with other embodiments without conflict.

[0144] The above embodiments merely illustrate several implementation methods of this application, and while the descriptions are relatively specific and detailed, they should not be construed as limiting the scope of patent protection. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application, and these all fall within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the appended claims.

Claims

1. A method of detecting a chassis component of a motor vehicle, characterized in that, include: Acquire images of target components containing the target chassis components of the vehicle to be inspected; Visual features of the target chassis component are extracted from the target component image; Obtain a pre-constructed fault feature database, which includes standard fault features and corresponding fault feature judgment conditions; The visual features are matched with standard fault features in the fault feature database to determine the standard fault features that match the visual features, and the corresponding fault feature judgment conditions are obtained based on the matching standard fault features. The visual features are compared with the fault feature judgment conditions, and the fault judgment result of the target chassis component is obtained based on the comparison result.

2. The method for testing motor vehicle chassis components according to claim 1, characterized in that, The method for constructing the fault feature database includes: Extract standard fault features from standard fault images of target chassis components; Transform fault knowledge in the field of vehicle chassis inspection into structured update rules; Based on the structured update rules, generate fault feature judgment conditions corresponding to the standard fault features; By associating the standard fault features with the corresponding fault feature judgment conditions, the fault feature database is obtained.

3. The method for testing motor vehicle chassis components according to claim 1, characterized in that, The step of extracting the visual features of the target chassis component based on the target component image includes: If the vehicle under inspection has historical component images containing the target chassis component, compare the historical component images and the target component images to obtain the difference areas between the historical component images and the target component images; Based on the difference regions in the target component image, the visual features of the target chassis component are extracted.

4. The method for testing motor vehicle chassis components according to claim 1, characterized in that, The step of acquiring the target component image containing the target chassis component of the vehicle to be inspected includes: Obtain the chassis image of the vehicle to be inspected; The chassis image is divided according to the target chassis component contained in the chassis image to obtain the target component image.

5. The method for testing motor vehicle chassis components according to claim 4, characterized in that, The step of acquiring the chassis image of the vehicle to be inspected includes: When the vehicle to be inspected enters the preset chassis image acquisition station, chassis images from different angles are obtained.

6. The method for testing motor vehicle chassis components according to claim 5, characterized in that, After obtaining chassis images from different angles, the method further includes: Preprocessing operations, including perspective correction, illumination equalization, and stitching together chassis images from different angles, yield a preprocessed chassis image.

7. The method for testing motor vehicle chassis components according to claim 1, characterized in that, After comparing the visual features with the fault feature judgment conditions and obtaining the fault judgment result of the target chassis component based on the comparison result, the method further includes: By integrating the fault determination results of multiple target chassis components contained in the chassis of the vehicle under inspection, a structured inspection report of the chassis of the vehicle under inspection is obtained.

8. A vehicle chassis component testing device, characterized in that, include: The module comprises an image acquisition module, a feature extraction module, and a fault determination module; among which: The image acquisition module is used to acquire images of target components, including target chassis components of the vehicle to be inspected. The feature extraction module is used to extract visual features of the target chassis component based on the target component image; The fault determination module is used to acquire a pre-constructed fault feature database, which includes standard fault features and corresponding fault feature judgment conditions; match the visual features with the standard fault features in the fault feature database to determine the standard fault features that match the visual features, and obtain the corresponding fault feature judgment conditions based on the matching standard fault features; compare the visual features with the fault feature judgment conditions, and obtain the fault determination result of the target chassis component based on the comparison result.

9. A vehicle chassis component testing system, characterized in that, The system includes: An image acquisition device is used to acquire images of the target components of the target chassis of the vehicle to be inspected. A component testing device for implementing the steps of the vehicle chassis component testing method according to any one of claims 1 to 7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the steps of the method for detecting motor vehicle chassis components as described in any one of claims 1 to 7.

Images