Method and system for detecting apparent performance of automobile bumper and medium

Abstract

The invention provides a method and a system for detecting the apparent performance of an automobile bumper and a medium. The method comprises the following steps: acquiring an apparent image of the automobile bumper; preprocessing the apparent image to obtain a preprocessed image; detecting the pre-processed image based on a preset detection model, and identifying the apparent defect of the automobile bumper in the pre-processed image to obtain an identification result; generating a detection report of the apparent performance of the automobile bumper according to the identification result; the apparent image is analyzed through the visual inspection technology, the apparent defect of the automobile bumper is accurately recognized according to the preset detection model, the defect type and the defect grade are obtained, and the detection precision is improved.

Description

[0001] This application relates to the field of automotive bumper appearance performance testing technology, and more specifically, to a method, system, and medium for testing automotive bumper appearance performance. Background Technology

[0002] As an important component of a vehicle's exterior, the appearance of a car bumper directly affects its visual quality, user experience, and market competitiveness. With the rapid development of the automotive industry, consumers have increasingly higher requirements for the surface smoothness, color consistency, and defect-free nature of bumpers. Therefore, efficient and accurate inspection of bumper appearance performance has become a key process in automotive manufacturing and quality control.

[0003] Currently, the appearance performance inspection of automotive bumpers mainly relies on manual visual inspection. Inspectors visually examine the bumper surface for defects such as scratches, dents, color differences, and bubbles, and determine the defect level based on experience. However, this method has significant limitations: firstly, manual inspection is inefficient and cannot meet the high-speed requirements of modern production lines, and prolonged inspections are prone to missed or false positives due to fatigue, resulting in poor consistency; secondly, defect judgment depends on subjective experience, and different inspectors have different standards for perceiving defects, leading to insufficient stability of inspection results and difficulty in achieving standardized quality control. Summary of the Invention

[0004] The purpose of this application is to provide a method, system, and medium for detecting the appearance performance of a car bumper. By analyzing the appearance image using visual inspection technology and accurately identifying the appearance defects of the car bumper according to a preset detection model, the method obtains the defect type and defect level, thereby improving the detection accuracy.

[0005] This application also provides a method for testing the appearance performance of an automobile bumper, including: Obtain a visual image of the car bumper; The apparent image is preprocessed to obtain a preprocessed image; The preprocessed image is detected based on a preset detection model to identify apparent defects in the car bumper in the preprocessed image, and the identification result is obtained. A test report on the appearance performance of the car bumper is generated based on the identification results.

[0006] Optionally, in the method for detecting the appearance performance of a car bumper described in the embodiments of this application, acquiring an appearance image of the car bumper specifically includes: The lighting conditions are set, including illumination from at least two different angles of light sources, with the angle between the light source and the surface of the bumper being 30°-60°. Under the set lighting conditions, multi-view images of the car bumper were acquired to obtain appearance images from multiple perspectives. Feature points are extracted from appearance images from multiple perspectives, and the overlapping areas of appearance images from adjacent perspectives are analyzed based on the feature point matching algorithm. Based on the coordinate transformation relationship of the overlapping areas, a weighted average fusion algorithm is used to stitch the images together to obtain a complete visual image of the car bumper.

[0007] Optionally, in the method for detecting the appearance performance of a car bumper described in this application embodiment, the appearance image is preprocessed to obtain a preprocessed image, specifically including: Obtain the appearance image and perform denoising processing on the appearance image based on Gaussian filtering algorithm, median filtering algorithm or bilateral filtering algorithm; The denoised appearance image is then converted to grayscale to obtain a grayscale image. The grayscale image is then subjected to contrast enhancement processing to obtain an enhanced image; The enhanced image is subjected to distortion correction based on the camera's intrinsic parameter matrix and distortion coefficients to obtain a preprocessed image.

[0008] Optionally, in the method for detecting the appearance performance of a car bumper described in this application embodiment, the preset detection model training method is as follows: Obtain a sample image set, wherein the sample images in the sample image set are labeled with the apparent defect type and defect location; The sample image set is divided into a training set and a validation set; An initial detection model is constructed based on a deep learning network; the initial detection model is trained using the training set. The model during the training process is validated based on the validation set to obtain validation results; Adjust the model parameters based on the verification results until the model converges to obtain the preset detection model.

[0009] Optionally, in the method for detecting the appearance performance of a car bumper described in this application embodiment, detecting the preprocessed image based on a preset detection model to identify appearance defects of the car bumper in the preprocessed image specifically includes: The preprocessed image is input into a preset detection model to obtain the confidence level of each candidate defect region; Candidate defect regions with a confidence level greater than a preset threshold are selected as the final identified apparent defects.

[0010] Optionally, in the method for detecting the appearance performance of a car bumper according to the embodiments of this application, after identifying the appearance defects of the car bumper in the preprocessed image, it further includes: The identified apparent defects were quantitatively analyzed to obtain the quantitative analysis results; The geometric parameters of the defect are determined based on the results of quantitative analysis, and the geometric parameters include at least one of length, width, area, and depth; The defect level of the apparent defect is obtained by comparing the geometric parameters with a preset defect level threshold.

[0011] Secondly, embodiments of this application provide a system for detecting the appearance performance of a car bumper. The system includes a memory and a processor. The memory includes a program for detecting the appearance performance of a car bumper. When the program for detecting the appearance performance of a car bumper is executed by the processor, it performs the following steps: Obtain a visual image of the car bumper; The apparent image is preprocessed to obtain a preprocessed image; The preprocessed image is detected based on a preset detection model to identify apparent defects in the car bumper in the preprocessed image, and the identification result is obtained. A test report on the appearance performance of the car bumper is generated based on the identification results.

[0012] Optionally, in the vehicle bumper appearance performance testing system described in this application embodiment, acquiring the appearance image of the vehicle bumper specifically includes: The lighting conditions are set, including illumination from at least two different angles of light sources, with the angle between the light source and the surface of the bumper being 30°-60°. Under the set lighting conditions, multi-view images of the car bumper were acquired to obtain appearance images from multiple perspectives. Feature points are extracted from appearance images from multiple perspectives, and the overlapping areas of appearance images from adjacent perspectives are analyzed based on the feature point matching algorithm. Based on the coordinate transformation relationship of the overlapping areas, a weighted average fusion algorithm is used to stitch the images together to obtain a complete visual image of the car bumper.

[0013] Optionally, in the vehicle bumper appearance performance detection system described in this application embodiment, the appearance image is preprocessed to obtain a preprocessed image, specifically including: Obtain the appearance image and perform denoising processing on the appearance image based on Gaussian filtering algorithm, median filtering algorithm or bilateral filtering algorithm; The denoised appearance image is then converted to grayscale to obtain a grayscale image. The grayscale image is then subjected to contrast enhancement processing to obtain an enhanced image; The enhanced image is subjected to distortion correction based on the camera's intrinsic parameter matrix and distortion coefficients to obtain a preprocessed image.

[0014] Thirdly, embodiments of this application also provide a computer-readable storage medium, which includes a method program for detecting the appearance performance of an automobile bumper. When the method program for detecting the appearance performance of an automobile bumper is executed by a processor, it implements the steps of the method for detecting the appearance performance of an automobile bumper as described in any of the preceding claims.

[0015] As can be seen from the above, the present application provides a method, system, and medium for detecting the appearance performance of a car bumper. This involves acquiring an image of the car bumper's appearance; preprocessing the image to obtain a preprocessed image; detecting the preprocessed image based on a preset detection model to identify appearance defects in the preprocessed image and obtain identification results; generating a report on the appearance performance of the car bumper based on the identification results; and analyzing the appearance image using visual inspection technology to accurately identify appearance defects in the car bumper according to the preset detection model, obtaining defect types and defect levels, thereby improving detection accuracy. Attached Figure Description

[0016] To more clearly illustrate the technical solutions of the embodiments of this application, the accompanying drawings used in the embodiments of this application will be briefly introduced below. It should be understood that the following drawings only show some embodiments of this application and should not be regarded as a limitation of the scope. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0017] Figure 1 A flowchart illustrating a method for testing the appearance performance of an automotive bumper provided in an embodiment of this application; Figure 2 A flowchart of a method for acquiring the appearance image of a car bumper, which is provided as an embodiment of this application for detecting the appearance performance of a car bumper; Figure 3 A flowchart of the appearance image preprocessing process for the method of detecting the appearance performance of an automobile bumper provided in this application embodiment. Detailed Implementation

[0018] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of the embodiments. The components of the embodiments of this application described and shown in the accompanying drawings can generally be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of this application provided in the accompanying drawings is not intended to limit the scope of the claimed application, but merely represents selected embodiments of this application. All other embodiments obtained by those skilled in the art based on the embodiments of this application without inventive effort are within the scope of protection of this application.

[0019] It should be noted that similar reference numerals and letters in the following figures indicate similar items; therefore, once an item is defined in one figure, it does not need to be further defined and explained in subsequent figures. Furthermore, in the description of this application, the terms "first," "second," etc., are used only to distinguish descriptions and should not be construed as indicating or implying relative importance.

[0020] Please refer to Figure 1 , Figure 1 This is a flowchart illustrating a method for testing the appearance performance of an automotive bumper, as described in some embodiments of this application. This method for testing the appearance performance of an automotive bumper is used in a terminal device and includes the following steps: S101, Obtain the appearance image of the car bumper; S102, preprocess the appearance image to obtain a preprocessed image; S103, Based on the preset detection model, the preprocessed image is detected to identify the apparent defects of the car bumper in the preprocessed image and obtain the identification result; S104, Generate an inspection report on the appearance performance of the car bumper based on the recognition results.

[0021] Please refer to Figure 2 , Figure 2 This is a flowchart illustrating a method for acquiring an appearance image of a car bumper, one of the embodiments of this application for detecting the appearance performance of a car bumper. According to embodiments of the present invention, acquiring an appearance image of a car bumper specifically includes: S201, Set lighting conditions, which include at least two different angles of light source illumination, with the angle between the light source and the bumper surface being 30°-60°; S202, under the set lighting conditions, multi-view images of the car bumper are acquired to obtain appearance images from multiple perspectives; S203, extract feature points from appearance images from multiple perspectives, and analyze the overlapping areas of appearance images from adjacent perspectives based on feature point matching algorithm; S204. Based on the coordinate transformation relationship of the overlapping areas, a weighted average fusion algorithm is used to stitch the images together to obtain a complete visual image of the car bumper.

[0022] Please refer to Figure 3 , Figure 3 This is a flowchart illustrating the appearance image preprocessing process of a method for detecting the appearance performance of an automobile bumper, as described in some embodiments of this application. According to embodiments of the present invention, preprocessing the appearance image to obtain a preprocessed image specifically includes: S301, acquire the appearance image, and perform noise reduction processing on the appearance image based on Gaussian filtering algorithm, median filtering algorithm or bilateral filtering algorithm; S302, perform grayscale processing on the denoised appearance image to obtain a grayscale image; S303, perform contrast enhancement processing on the grayscale image to obtain the enhanced image; S304 performs distortion correction on the enhanced image based on the camera's intrinsic parameter matrix and distortion coefficients to obtain a preprocessed image.

[0023] It should be noted that Gaussian filtering: by applying a weighted average to the area surrounding each pixel in an image (the weights follow a Gaussian distribution), it smooths image details and effectively suppresses Gaussian noise (such as sensor thermal noise), but may slightly blur edge information. It is suitable for scenarios with relatively uniform noise distribution.

[0024] Median filtering: Replaces the pixel value with the median value in the neighborhood of the pixel. It has a significant effect on suppressing impulse noise (such as salt and pepper noise, which may be caused by momentary interference from the device) and can preserve edge details well. It is suitable for processing bumper images containing isolated noise points (such as bright spots caused by surface dust reflection).

[0025] Bilateral filtering combines spatial domain Gaussian weights and gray-level similarity weights to smooth noise while preserving image edges (such as scratches and bubble edges), avoiding edge blurring caused by traditional filtering. It is especially suitable for preprocessing complex textures or subtle defects on the surface of bumpers.

[0026] Selection criteria: The algorithm is dynamically selected based on the actual noise type. If the image is mainly composed of granular Gaussian noise, Gaussian filtering is preferred; if it contains obvious isolated noise points, median filtering is selected; if it is necessary to preserve the edge details of defects, bilateral filtering is used.

[0027] Converting a color image (RGB three channels) into a single-channel grayscale image reduces data dimensionality and computational load in subsequent processing, while unifying the brightness characteristics of the image to provide a more stable input for defect detection.

[0028] Bumpers are mostly curved structures, and when capturing images, they are prone to local areas that are too dark or too bright due to differences in lighting angles (such as shadows in recesses or reflections on smooth surfaces), which compresses the grayscale difference between defects and the background (for example, shallow scratches may be submerged in low-contrast areas).

[0029] Commonly used augmentation algorithms include: Global histogram equalization: Improves overall contrast by stretching the range of grayscale values, but may amplify noise or over-enhance background texture.

[0030] Adaptive Histogram Equalization (CLAHE): Divides the image into multiple sub-regions, performs histogram equalization on each sub-region separately, and avoids noise amplification by limiting contrast. It is suitable for handling scenes with uneven lighting on the surface of bumpers and can specifically enhance local defects (such as grayscale changes at recessed edges).

[0031] In practical applications, adaptive histogram equalization is preferred to ensure that defects are enhanced without compromising the overall stability of the image.

[0032] When industrial cameras capture images, the lens produces distortion due to optical refraction deviations: radial distortion manifests as "barrel-shaped" or "pincushion-shaped" deformations in the image edge areas (such as scratches on the edge of a bumper being stretched or compressed); tangential distortion manifests as image tilting (such as planar defects being distorted into curves). These distortions can lead to measurement errors in the geometric parameters of defects (such as length and area), and even misidentification of defect types (such as misidentifying straight scratches as curves).

[0033] The correction process relies on the intrinsic parameter matrix (including focal length, principal point coordinates, etc.) and distortion coefficients (describing the degree of radial / tangential distortion) obtained from camera calibration. The pixel coordinates are reverse-mapped through a mathematical model (such as the Brownian distortion model) to correct the position of distorted pixels, so that the bumper surface in the image can be restored to its true proportions and shape.

[0034] After correction, the preprocessed image can accurately reflect the geometric features of the bumper, providing a reliable image basis for subsequent quantitative analysis of defects (such as scratch length measurement and dent depth calculation).

[0035] According to an embodiment of the present invention, the training method for the preset detection model is as follows: Obtain a sample image set, in which the sample images are labeled with the apparent defect type and defect location; The sample image set is divided into a training set and a validation set; An initial detection model is built based on a deep learning network; the initial detection model is trained using a training set; The model during the training process is validated based on the validation set to obtain the validation results; Adjust the model parameters based on the verification results until the model converges to obtain the preset detection model.

[0036] It should be noted that the following parts need to be analyzed during the training of the preset detection model: Composition of the sample image set: The samples need to cover typical scenarios of car bumpers, including: Defect diversity: Includes all target detection defect types such as scratches (length, depth), dents (size, depth), color differences (brightness, range), and bubbles (number, diameter); Scene diversity: Covers bumper images of different materials (polypropylene, ABS resin, etc.), colors (black, white, silver, etc.), lighting conditions (light sources from different angles), and shooting perspectives, enhancing the model's generalization ability; Scale of images: A certain scale (usually thousands to tens of thousands) is required to avoid model overfitting (only remembering training samples and being unable to recognize new images).

[0037] Annotation content: Sample images are manually or semi-automatically annotated using professional annotation tools (such as LabelImg, VGG Image Annotator), including: Defect type: Label each defect with a category (e.g., "scratches" or "dents"); Defect location: Mark the specific area of ​​the defect in the image using bounding boxes (such as the coordinates of the top left corner (x1, y1) and the bottom right corner (x2, y2)); Labeling accuracy: Ensures that the bounding box accurately surrounds the defect, the type label is unambiguous, and provides a reliable learning target for the model.

[0038] The sample image set is divided into a training set and a validation set, as follows: Split ratio: The training set and validation set are usually split in a ratio of 7:3, 8:2 or 9:1 (e.g., out of 10,000 samples, 7,000 are the training set and 3,000 are the validation set). The training set needs to be large enough to support model learning, and the validation set needs to be representative to reflect the true performance of the model.

[0039] Divide the data into two sets: random sampling is used to ensure that the defect types and distribution characteristics of the training set and the validation set are consistent (e.g., the proportion of scratch defects in both sets is similar) to avoid distortion of validation results due to differences in data distribution.

[0040] Validation metrics: The core performance metrics of the model are calculated using a validation set, including: Accuracy: The proportion of actual defects in the areas predicted as defects (reflecting the ability to resist false alarms). Recall: The proportion of all real defects that are correctly identified by the model (reflecting the ability to resist missed detections). Average precision (mAP): A combined metric of precision and recall at different thresholds, measuring the overall performance of the model.

[0041] Validation frequency: After a certain number of training epochs (e.g., 1 epoch), evaluate the model using a validation set and record the performance change curve (e.g., mAP changes with the number of iterations). If the validation set accuracy is low (many false positives): it may be that the model is overly focused on noisy features. It is necessary to reduce the learning rate, increase regularization (such as L2 regularization), or expand the training samples. If the validation set recall is low (many missed detections): it may be that the model has not fully learned the defect features. It is necessary to increase the number of iterations, adjust the network structure (such as deepening the convolutional layers), or add difficult samples (such as shallow scratches, small bubbles). If the validation performance fluctuates greatly, it may be due to an unreasonable batch size setting. You need to adjust the batch size or add data augmentation (such as rotating or scaling the samples).

[0042] Convergence determination: When the mAP and other metrics of the validation set no longer improve significantly for multiple consecutive rounds (e.g., 5-10 epochs) (the change is less than the preset threshold, e.g., 0.1%), and the performance difference between the training set and the validation set is small (no obvious overfitting), the model is determined to have converged, and the model parameters at this time are saved as the preset detection model.

[0043] According to an embodiment of the present invention, a preprocessed image is detected based on a preset detection model to identify apparent defects in the car bumper in the preprocessed image, specifically including: The preprocessed image is input into a preset detection model to obtain the confidence level of each candidate defect region; Candidate defect regions with a confidence level greater than a preset threshold are selected as the final identified apparent defects.

[0044] It should be noted that the preset detection model is built upon deep learning networks (such as YOLO, Faster R-CNN, SSD, etc.) and, after training with a large number of labeled samples, possesses the ability to automatically identify defects from images. Its core functions include: Feature extraction: Extract key visual features of defects from the preprocessed image (such as the edge continuity of scratches, gray-level gradient changes of depressions, color distribution differences of color differences, circular outlines and internal brightness of bubbles, etc.). Defect localization: Predicting the location of defects in an image (usually represented by bounding box coordinates, such as the pixel coordinates of the top left and bottom right corners); Type classification: Determine the type of candidate defect (such as scratches, dents, etc.); Confidence output: For each predicted candidate defect region, output a confidence value between 0 and 1, representing the model's confidence level that the region "belongs to a certain type of defect" (the closer the value is to 1, the higher the confidence).

[0045] Candidate defect region generation: The model performs a global scan of the preprocessed image and predicts all regions that may contain defects. For example, for a bumper image containing scratches and bubbles, the model may output multiple candidate regions, each corresponding to a predicted defect type (such as "scratch" or "bubble") and a corresponding confidence level (such as 0.92, 0.85, etc.).

[0046] Candidate defect regions with confidence levels greater than a preset threshold are selected as the final identified apparent defects. False alarm regions with low confidence levels in the model prediction (such as misclassifying normal textures as defects) are eliminated to improve the accuracy of defect identification and ensure the reliability of the output results.

[0047] After threshold screening, the remaining candidate defect areas are the final identified surface defects, including information such as defect type (e.g., "dent"), location (boundary box coordinates) and confidence level (e.g., 0.78), which can be directly used for subsequent quantitative analysis (e.g., size measurement, grade determination).

[0048] According to an embodiment of the present invention, after identifying apparent defects in a car bumper in a preprocessed image, the method further includes: The identified apparent defects were quantitatively analyzed to obtain the quantitative analysis results; The geometric parameters of the defect are determined based on the results of quantitative analysis. The geometric parameters include at least one of length, width, area, and depth. The defect level of the apparent defect is obtained by comparing the geometric parameters with the preset defect level threshold.

[0049] According to an embodiment of the present invention, the preset defect level threshold includes a first-level threshold, a second-level threshold, and a third-level threshold, which correspond to minor defects, moderate defects, and severe defects, respectively. The level of apparent defects is determined based on the comparison results between geometric parameters and preset defect level thresholds. When the geometric parameters are less than the first-level threshold, it is judged as a minor defect; When the geometric parameters are greater than or equal to the first-level threshold and less than the second-level threshold, it is judged as a moderate defect; When the geometric parameters are greater than or equal to the secondary threshold, it is judged as a serious defect.

[0050] It should be noted that by setting multiple threshold levels to compare geometric parameters, the defect accuracy of apparent defects can be determined, thereby achieving graded defect detection and improving detection accuracy.

[0051] Secondly, embodiments of this application provide a system for detecting the appearance performance of a car bumper. The system includes a memory and a processor. The memory contains a program for detecting the appearance performance of a car bumper. When the program for detecting the appearance performance of a car bumper is executed by the processor, it performs the following steps: Obtain a visual image of the car bumper; The appearance image is preprocessed to obtain a preprocessed image; The preprocessed image is detected based on a preset detection model to identify apparent defects in the car bumper and obtain the identification results. A test report on the appearance performance of the car bumper is generated based on the identification results.

[0052] According to an embodiment of the present invention, obtaining an apparent image of a car bumper specifically includes: Set the lighting conditions, which include at least two different angles of light source illumination, with the angle between the light source and the bumper surface being 30°-60°. Under the set lighting conditions, multi-view images of the car bumper were acquired to obtain appearance images from multiple perspectives. Feature points are extracted from appearance images from multiple perspectives, and the overlapping areas of appearance images from adjacent perspectives are analyzed based on the feature point matching algorithm. Based on the coordinate transformation relationship of the overlapping areas, a weighted average fusion algorithm is used to stitch the images together to obtain a complete visual image of the car bumper.

[0053] According to an embodiment of the present invention, preprocessing is performed on the apparent image to obtain a preprocessed image, specifically including: Obtain the appearance image and perform noise reduction on the appearance image based on Gaussian filtering algorithm, median filtering algorithm, or bilateral filtering algorithm; The denoised appearance image is then converted to grayscale to obtain a grayscale image. A grayscale image is contrast-enhanced to obtain an enhanced image; The enhanced image is subjected to distortion correction based on the camera's intrinsic parameter matrix and distortion coefficients to obtain a preprocessed image.

[0054] A third aspect of the present invention provides a computer-readable storage medium including a method program for detecting the appearance performance of an automobile bumper. When the method program for detecting the appearance performance of an automobile bumper is executed by a processor, it implements the steps of the method for detecting the appearance performance of an automobile bumper as described in any of the above claims.

[0055] This invention discloses a method, system, and medium for detecting the appearance performance of a car bumper. The method involves acquiring an image of the car bumper's appearance; preprocessing the image to obtain a preprocessed image; detecting the preprocessed image based on a preset detection model to identify appearance defects in the car bumper and obtain identification results; generating a report on the appearance performance of the car bumper based on the identification results; and analyzing the appearance image using visual inspection technology and accurately identifying appearance defects in the car bumper according to the preset detection model, thereby obtaining the defect type and defect level and improving detection accuracy.

[0056] In the several embodiments provided in this application, it should be understood that the disclosed devices and methods can be implemented in other ways. The device embodiments described above are merely illustrative. For example, the division of units is only a logical functional division, and in actual implementation, there may be other division methods, such as: multiple units or components can be combined, or integrated into another system, or some features can be ignored or not executed. In addition, the coupling, direct coupling, or communication connection between the various components shown or discussed can be through some interfaces, and the indirect coupling or communication connection between devices or units can be electrical, mechanical, or other forms.

[0057] The units described above as separate components may or may not be physically separate. The components shown as units may or may not be physical units. They may be located in one place or distributed across multiple network units. Some or all of the units may be selected to achieve the purpose of this embodiment according to actual needs.

[0058] In addition, in the various embodiments of the present invention, each functional unit can be integrated into one processing unit, or each unit can be a separate unit, or two or more units can be integrated into one unit; the integrated unit can be implemented in hardware or in the form of hardware plus software functional units.

[0059] Those skilled in the art will understand that all or part of the steps of the above method embodiments can be implemented by hardware related to program instructions. The aforementioned program can be stored in a readable storage medium. When the program is executed, it performs the steps of the above method embodiments. The aforementioned storage medium includes various media that can store program code, such as mobile storage devices, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0060] Alternatively, if the integrated units of this invention are implemented as software functional modules and sold or used as independent products, they can also be stored in a readable storage medium. Based on this understanding, the technical solutions of the embodiments of this invention, or the parts that contribute to the prior art, can be embodied in the form of a software product. This software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the methods described in the various embodiments of this invention. The aforementioned storage medium includes various media capable of storing program code, such as mobile storage devices, ROM, RAM, magnetic disks, or optical disks.

Claims

1. A method for detecting the apparent performance of an automobile bumper, characterized by, The method comprises the following steps: obtaining an apparent image of an automobile bumper; preprocessing the apparent image to obtain a preprocessed image; detecting the preprocessed image based on a preset detection model to identify the apparent defects of the automobile bumper in the preprocessed image and obtain an identification result; generating a detection report of the apparent performance of the automobile bumper according to the identification result.

2. The method of claim 1, wherein the method further comprises: The method for obtaining an apparent image of an automobile bumper comprises the following steps: setting illumination conditions, wherein the illumination conditions comprise at least two different angle light source illuminations, and the included angle between the light source and the surface of the bumper is 30°-60°; under the set illumination conditions, performing multi-view image acquisition on the automobile bumper to obtain apparent images of multiple views; extracting feature points of the apparent images of multiple views, and analyzing the overlapping areas of the apparent images of adjacent views based on a feature point matching algorithm; according to the coordinate transformation relationship of the overlapping areas, performing splicing processing on the images by using a weighted average fusion algorithm to obtain a complete apparent image of the automobile bumper.

3. The method of claim 2, wherein the step of determining the presence of a defect comprises the steps of: determining the presence of a defect if the value of the parameter is greater than a predetermined threshold value. The method for preprocessing the apparent image to obtain a preprocessed image comprises the following steps: obtaining the apparent image, and performing denoising processing on the apparent image based on a Gaussian filtering algorithm, a median filtering algorithm or a bilateral filtering algorithm; performing grayscale processing on the denoised apparent image to obtain a grayscale image; performing contrast enhancement processing on the grayscale image to obtain an enhanced image; performing distortion correction processing on the enhanced image based on the intrinsic matrix and distortion coefficient of the camera to obtain the preprocessed image.

4. The method of claim 3, wherein the step of determining the presence of a defect comprises the steps of: determining the presence of a defect if the value of the parameter is greater than a predetermined threshold value. The preset detection model training method comprises the following steps: obtaining a sample image set, wherein the sample images in the sample image set are labeled with apparent defect types and defect positions; dividing the sample image set into a training set and a validation set; constructing an initial detection model based on a deep learning network, training the initial detection model using the training set; verifying the model during the training process based on the validation set to obtain a verification result; adjusting the model parameters according to the verification result until the model converges to obtain the preset detection model.

5. The method of claim 4, wherein the step of determining the presence of a defect comprises the steps of: determining the presence of a defect if the value of the parameter is greater than a predetermined threshold value. The method for detecting the preprocessed image based on the preset detection model to identify the apparent defects of the automobile bumper in the preprocessed image comprises the following steps: inputting the preprocessed image into the preset detection model to obtain the confidence of each candidate defect region; screening out the candidate defect regions with a confidence greater than a preset threshold as the final identified apparent defects.

6. The method of claim 5, wherein the step of determining the presence of a defect comprises the steps of: determining the presence of a defect if the value of the parameter is greater than a predetermined threshold value. After identifying the apparent defects of the automobile bumper in the preprocessed image, the method further comprises the following steps: performing quantitative analysis on the identified apparent defects to obtain a quantitative analysis result; determining the geometric parameters of the defects based on the quantitative analysis result, wherein the geometric parameters comprise at least one of length, width, area and depth; comparing the geometric parameters with preset defect level thresholds to obtain the defect level of the apparent defects.

7. A system for detecting the apparent performance of an automobile bumper, characterized by The system comprises a memory and a processor, the memory comprises a program of a method for detecting the apparent performance of an automobile bumper, and the program of the method for detecting the apparent performance of the automobile bumper is executed by the processor to implement the following steps: obtaining an apparent image of an automobile bumper; preprocessing the apparent image to obtain a preprocessed image; Detect the preprocessed image based on a preset detection model, identify the apparent defects of the bumper of the vehicle in the preprocessed image, and obtain an identification result; Generate a detection report of the apparent performance of the bumper of the vehicle according to the identification result.

8. The system for detecting the apparent performance of an automobile bumper according to claim 7, wherein An apparent image of the bumper of the vehicle is acquired, specifically including: Setting illumination conditions, the illumination conditions including illumination of at least two light sources at different angles, the included angle between the light source and the surface of the bumper being 30°-60°; Under the set illumination conditions, multi-view image acquisition is performed on the bumper of the vehicle to obtain apparent images at multiple views; Feature points of the apparent images at multiple views are extracted, and the overlapping regions of the apparent images at adjacent views are analyzed based on a feature point matching algorithm; According to the coordinate transformation relationship of the overlapping regions, a weighted average fusion algorithm is used to perform splicing processing on the images to obtain a complete apparent image of the bumper of the vehicle.

9. The system for detecting the apparent performance of an automobile bumper according to claim 8, wherein The apparent image is preprocessed to obtain a preprocessed image, specifically including: An apparent image is acquired, and a denoising process is performed on the apparent image based on a Gaussian filter algorithm, a median filter algorithm or a bilateral filter algorithm; The denoised apparent image is subjected to grayscale processing to obtain a grayscale image; The grayscale image is subjected to contrast enhancement processing to obtain an enhanced image; The enhanced image is subjected to distortion correction processing based on the intrinsic matrix and distortion coefficients of the camera to obtain a preprocessed image.

10. A computer-readable storage medium, characterized in that, The computer readable storage medium includes a bumper of the vehicle apparent performance detection method program, the bumper of the vehicle apparent performance detection method program is executed by the processor, realizes the steps of the bumper of the vehicle apparent performance detection method as claimed in any one of claims 1 to 6.

Images