A method and system for detecting vehicle tire damage

Abstract

This invention discloses a method and system for detecting vehicle tire damage. In this invention, multiple cameras are used to continuously photograph the tires of a moving vehicle, obtaining partial images of the tires at different locations. These partial images are then stitched together to form a complete tire image. By comparing the tire image with a pre-stored grayscale tire image, the damage location can be directly and accurately located. This solves the problem that existing methods for detecting vehicle tire faults using neural networks have complex models and low detection accuracy when there are insufficient samples for training.

Description

Technical Field

[0001] This invention relates to the field of image processing technology, and specifically to a method and system for detecting vehicle tire damage. Background Technology

[0002] Intelligent rail transit vehicles are public transportation, and their wheels need to be inspected regularly for abnormal wear or damage. However, visual inspection of the wheel surface is difficult due to the limited space, making it hard for personnel to observe the entire surface of the wheel. This may result in missed or forgotten inspections, causing wheel problems to spread to the operation site and affecting public transportation safety.

[0003] Existing methods for vehicle tire fault detection using neural networks can overcome the drawbacks of manual inspection, but they suffer from complex models and low detection accuracy when there are insufficient samples for training. Summary of the Invention

[0004] To address the aforementioned shortcomings in the existing technology, the present invention provides a vehicle tire damage detection method and system that solves the problem that existing vehicle tire fault detection methods using neural networks have complex models and low detection accuracy when there are insufficient samples for training.

[0005] To achieve the above-mentioned objectives, the technical solution adopted by this invention is as follows: a method for detecting vehicle tire damage, comprising the following steps:

[0006] S1. Acquire local images of vehicle tires at different locations;

[0007] S2. Stitch together multiple partial images of the vehicle tires from different locations to obtain a vehicle tire image;

[0008] S3. Compare the vehicle tire image with the pre-stored grayscale vehicle tire image to obtain the location of the damage.

[0009] Furthermore, step S1 includes the following sub-steps:

[0010] S11. Fix multiple cameras at the same intervals underground;

[0011] S12. When the vehicle tires pass by, a camera is used to take pictures of the vehicle tires to obtain partial images of the vehicle tires at different positions.

[0012] Furthermore, step S2 includes the following sub-steps:

[0013] S21. Based on the optical characteristics of the camera, perform distortion correction on each local image of the vehicle tire to obtain a corrected image;

[0014] S22. Remove the overlapping parts of each corrected image during stitching to obtain the stitched block;

[0015] S23. Assemble each stitching block in the order of the camera to obtain the vehicle tire image.

[0016] Furthermore, in step S22, the length of the overlapping portion removed during stitching of each corrected image is:

[0017]

[0018] Where d is the length of the images removed from both sides of each corrected image, N is the number of cameras, L is the outer circumference of the vehicle tire in the image, S is the length of the tire in the image captured by each camera, and γ is the imaging coefficient.

[0019] Furthermore, step S3 includes the following sub-steps:

[0020] S31. Process the vehicle tire image into grayscale to obtain a grayscale image;

[0021] S32. Remove the background portion from the grayscale image to obtain the tire image;

[0022] S33. Adjust the brightness of the tire image to obtain a standard brightness tire image;

[0023] S34. Compare the standard brightness tire image with the pre-stored grayscale vehicle tire image to obtain the location of the damage.

[0024] Furthermore, step S32 includes the following sub-steps:

[0025] S321. Extract edge pixels of the tire on one side or the opposite side of the grayscale image;

[0026] S322. Calculate the distance between the edge pixel and the surrounding pixels to obtain the grayscale distance;

[0027] S323. Determine whether the grayscale distance is less than the distance threshold. If yes, the surrounding pixels are edge pixels, and jump to step S324. If no, the surrounding pixels are tire pixels, and jump to step S324.

[0028] S324. Determine whether all edge pixels in the grayscale image have been selected. If yes, proceed to step S325. If no, select the unselected edge pixels and proceed to step S322.

[0029] S325. Remove all edge pixels from the grayscale image to obtain the tire image.

[0030] The beneficial effects of the above-mentioned further solution are as follows: In this invention, the edge pixels on both sides and the edge pixels found around them all need to perform steps S322 to S325. When the grayscale difference is not significant, they are edge pixels. Check the pixels around each edge pixel to find all the edge pixels. Then the edge pixels can be removed from the grayscale image, and the remaining pixels are tire pixels.

[0031] Furthermore, step S33 includes the following sub-steps:

[0032] S331. Take multiple edge pixels of the pre-stored grayscale vehicle tire image and obtain the grayscale values ​​of the multiple pre-stored edge pixels.

[0033] S332. Take multiple edge pixels at the same position in the tire image to obtain the grayscale values ​​of multiple tire edge pixels;

[0034] S333. Calculate the brightness coefficient based on the gray values ​​of multiple pre-stored edge pixels and multiple tire edge pixels;

[0035] S334. Adjust the brightness of the tire pattern according to the brightness coefficient to obtain a standard brightness tire pattern.

[0036] The beneficial effect of the above-mentioned further solution is that the brightness coefficient is calculated by taking the gray values ​​of pixels at the same position, and the brightness of the tire image is adjusted by the brightness coefficient to eliminate the influence of light intensity on the image.

[0037] Furthermore, the formula for calculating the brightness coefficient in step S333 is as follows:

[0038]

[0039] Where δ is the brightness coefficient, N is the number of pre-stored edge pixels or the number of tire edge pixels, and G′ i Let G″ be the grayscale value of the i-th tire edge pixel. i Let i be the grayscale value of the i-th pre-stored edge pixel;

[0040] The formula for adjusting the brightness of the tire map in step S334 is as follows:

[0041]

[0042] Among them, g i G represents the grayscale value of the j-th pixel on the standard brightness tire map. j Let be the grayscale value of the j-th pixel in the tire image, and M be the number of pixels in the tire image.

[0043] Furthermore, step S34 includes the following sub-steps:

[0044] S341. Divide the standard brightness tire image into multiple tire image blocks;

[0045] S342. Divide the pre-stored grayscale vehicle tire image into multiple pre-stored image blocks using the same division method;

[0046] S343. Calculate the total grayscale value for each tire image block and the pre-stored image block;

[0047] S345. Subtract the total gray value of the pre-stored image block at the same position from the total gray value of the tire image block to obtain the total difference.

[0048] S346. Select tire image blocks whose total difference exceeds the total threshold as damage locations.

[0049] The beneficial effects of the above-mentioned further scheme are as follows: the tire image block and the pre-stored image block are divided in the same way, and the abnormal image block can be found by simply using the difference in gray level between each pair of image blocks, thereby determining the location of the damage.

[0050] A system for detecting vehicle tire damage includes: a camera, an image stitching unit, and a loss recognition unit;

[0051] The camera is used to capture partial images of vehicle tires at different locations;

[0052] The image stitching unit is used to stitch together multiple partial images of vehicle tires at different locations to obtain a vehicle tire image.

[0053] The loss recognition unit is used to compare the vehicle tire image with a pre-stored grayscale vehicle tire image to obtain the damage location.

[0054] In summary, the beneficial effects of this invention are as follows: This invention uses multiple cameras to continuously photograph the tires of a moving vehicle, obtaining partial images of the tires at different locations. These partial images are then stitched together to form a complete tire image. By comparing the tire image with a pre-stored grayscale tire image, the location of the damage can be directly and accurately identified. This solves the problem that existing methods for tire fault detection using neural networks suffer from complex models and low detection accuracy when there are insufficient samples for training. Attached Figure Description

[0055] Figure 1 A flowchart of a method for detecting vehicle tire damage;

[0056] Figure 2 This is a diagram showing the installation of the camera.

[0057] Figure 3 This is a schematic diagram illustrating the principle of image stitching. Detailed Implementation

[0058] The specific embodiments of the present invention are described below to enable those skilled in the art to understand the present invention. However, it should be understood that the present invention is not limited to the scope of the specific embodiments. For those skilled in the art, various changes are obvious as long as they are within the spirit and scope of the present invention as defined and determined by the appended claims. All inventions utilizing the concept of the present invention are protected.

[0059] like Figure 1 As shown, a method for detecting vehicle tire damage includes the following steps:

[0060] S1. Acquire local images of vehicle tires at different locations;

[0061] like Figure 2 As shown, step S1 includes the following sub-steps:

[0062] S11. Fix multiple cameras at the same intervals underground;

[0063] S12. When the vehicle tires pass by, a camera is used to take pictures of the vehicle tires to obtain partial images of the vehicle tires at different positions.

[0064] S2. Stitch together multiple partial images of the vehicle tires from different locations to obtain a vehicle tire image;

[0065] Step S2 includes the following sub-steps:

[0066] S21. Based on the optical characteristics of the camera, perform distortion correction on each local image of the vehicle tire to obtain a corrected image;

[0067] S22. Remove the overlapping parts of each corrected image during stitching to obtain the stitched block;

[0068] like Figure 3 As shown, in step S22, the length of the overlapping portion removed during stitching of each corrected image is:

[0069]

[0070] Where d is the length of the images removed from both sides of each corrected image, N is the number of cameras, L is the outer circumference of the vehicle tire in the image, S is the length of the tire in the image captured by each camera, and γ is the imaging coefficient.

[0071] S23. Assemble each stitching block in the order of the camera to obtain the vehicle tire image.

[0072] S3. Compare the vehicle tire image with the pre-stored grayscale vehicle tire image to obtain the location of the damage.

[0073] Step S3 includes the following sub-steps:

[0074] S31. Process the vehicle tire image into grayscale to obtain a grayscale image;

[0075] S32. Remove the background portion from the grayscale image to obtain the tire image;

[0076] Step S32 includes the following sub-steps:

[0077] S321. Extract edge pixels of the tire on one side or the opposite side of the grayscale image;

[0078] In this embodiment, step S321 only considers the case where there are edge pixels on both sides. If there are edge pixels on all four sides of the image, then the edge pixels need to be found and removed according to steps S322 to S325.

[0079] S322. Calculate the distance between the edge pixel and the surrounding pixels to obtain the grayscale distance;

[0080] In step S322, the grayscale distance is the difference in grayscale values ​​between pixels.

[0081] S323. Determine whether the grayscale distance is less than the distance threshold. If yes, the surrounding pixels are edge pixels, and jump to step S324. If no, the surrounding pixels are tire pixels, and jump to step S324.

[0082] S324. Determine whether all edge pixels in the grayscale image have been selected. If yes, proceed to step S325. If no, select the unselected edge pixels and proceed to step S322.

[0083] S325. Remove all edge pixels from the grayscale image to obtain the tire image.

[0084] In this invention, the edge pixels on both sides and the edge pixels found around them all need to be processed through steps S322 to S325. When the grayscale difference is not significant, they are edge pixels. The pixels around each edge pixel are checked to find all the edge pixels. Then the edge pixels can be removed from the grayscale image, and the remaining pixels are tire pixels.

[0085] S33. Adjust the brightness of the tire image to obtain a standard brightness tire image;

[0086] Step S33 includes the following sub-steps:

[0087] S331. Take multiple edge pixels of the pre-stored grayscale vehicle tire image and obtain the grayscale values ​​of the multiple pre-stored edge pixels.

[0088] S332. Take multiple edge pixels at the same position in the tire image to obtain the grayscale values ​​of multiple tire edge pixels;

[0089] S333. Calculate the brightness coefficient based on the gray values ​​of multiple pre-stored edge pixels and multiple tire edge pixels;

[0090] The formula for calculating the luminance coefficient in step S333 is as follows:

[0091]

[0092] Where δ is the brightness coefficient, N is the number of pre-stored edge pixels or the number of tire edge pixels, and G′ i Let G″ be the grayscale value of the i-th tire edge pixel. i Let i be the grayscale value of the i-th pre-stored edge pixel;

[0093] S334. Adjust the brightness of the tire pattern according to the brightness coefficient to obtain a standard brightness tire pattern.

[0094] The formula for adjusting the brightness of the tire map in step S334 is as follows:

[0095]

[0096] Among them, g i G represents the grayscale value of the j-th pixel on the standard brightness tire map. j Let be the grayscale value of the j-th pixel in the tire image, and M be the number of pixels in the tire image.

[0097] The brightness coefficient is calculated by taking the gray values ​​of pixels at the same location. The brightness coefficient is then used to adjust the brightness of the tire image, eliminating the influence of light intensity on the image.

[0098] S34. Compare the standard brightness tire image with the pre-stored grayscale vehicle tire image to obtain the location of the damage.

[0099] Step S34 includes the following sub-steps:

[0100] S341. Divide the standard brightness tire image into multiple tire image blocks;

[0101] S342. Divide the pre-stored grayscale vehicle tire image into multiple pre-stored image blocks using the same division method;

[0102] S343. Calculate the total grayscale value for each tire image block and the pre-stored image block;

[0103] In step S343, the total grayscale value is the sum of the grayscale values ​​of all pixels.

[0104] S345. Subtract the total gray value of the pre-stored image block at the same position from the total gray value of the tire image block to obtain the total difference.

[0105] S346. Select tire image blocks whose total difference exceeds the total threshold as damage locations.

[0106] The tire image blocks and the pre-stored image blocks are divided in the same way. Abnormal image blocks can be found by simply using the difference in gray levels between each pair of image blocks, thereby determining the location of the damage.

[0107] A system for detecting vehicle tire damage includes: a camera, an image stitching unit, and a loss recognition unit;

[0108] The camera is used to capture partial images of vehicle tires at different locations;

[0109] The image stitching unit is used to stitch together multiple partial images of vehicle tires at different locations to obtain a vehicle tire image.

[0110] The loss recognition unit is used to compare the vehicle tire image with a pre-stored grayscale vehicle tire image to obtain the damage location.

[0111] In summary, the beneficial effects of this invention are as follows: This invention uses multiple cameras to continuously photograph the tires of a moving vehicle, obtaining partial images of the tires at different locations. These partial images are then stitched together to form a complete tire image. By comparing the tire image with a pre-stored grayscale tire image, the location of the damage can be directly and accurately identified. This solves the problem that existing methods for tire fault detection using neural networks suffer from complex models and low detection accuracy when there are insufficient samples for training.

Claims

1. A method of detecting damage to a vehicle tire, characterized by, The method comprises the following steps: S1, collecting vehicle tire local images at different positions; S2, splicing multiple vehicle tire local images at different positions to obtain a vehicle tire image; S3, comparing the vehicle tire image with a pre-stored gray vehicle tire image to obtain a damage position; The step S3 comprises the following steps: S31, performing gray processing on the vehicle tire image to obtain a gray image; S32, removing the background part in the gray image to obtain a tire image; S33, adjusting the brightness of the tire image to obtain a standard brightness tire image; S34, comparing the standard brightness tire image with the pre-stored gray vehicle tire image to obtain a damage position; The step S32 comprises the following steps: S321, taking the edge pixel points on one side or the other side of the tire in the gray image; S322, calculating the distance between the edge pixel points and the surrounding pixel points to obtain a gray distance; S323, judging whether the gray distance is less than a distance threshold value, if yes, the surrounding pixel points are edge pixel points, and jumping to step S324, if not, the surrounding pixel points are tire pixel points, and jumping to step S324; S324, judging whether the edge pixel points in the gray image are taken, if yes, jumping to step S325, if not, taking the edge pixel points that are not selected, and jumping to step S322; S325, removing all the edge pixel points from the gray image to obtain a tire image; The step S33 comprises the following steps: S331, taking multiple edge pixel points of the pre-stored gray vehicle tire image to obtain the gray values of the multiple pre-stored edge pixel points; S332, taking multiple edge pixel points at the same position of the tire image to obtain the gray values of the multiple tire edge pixel points; S333, calculating a brightness coefficient according to the gray values of the multiple pre-stored edge pixel points and the gray values of the multiple tire edge pixel points; S334, adjusting the brightness of the tire image according to the brightness coefficient to obtain a standard brightness tire image; The calculation formula of the brightness coefficient in the step S333 is: wherein, is a luminance coefficient, is a number of pre-stored edge pixel points or a number of tire edge pixel points, is a gray value of the th tire edge pixel point, is a gray value of the th pre-stored edge pixel point; The formula for adjusting the brightness of the tire image in the step S334 is: wherein, is the gray value of the i-th pixel point on the standard brightness tire image, is the gray value of the i-th pixel point on the standard brightness tire image, is the gray value of the i-th pixel point on the standard brightness tire image, is the gray value of the i-th pixel point on the standard brightness tire image, is the number of pixel points of the tire image.

2. The method of claim 1, wherein The step S1 comprises the following steps: S11, fixing multiple cameras at the same interval on the ground; S12, when the vehicle tire passes, using the cameras to shoot the vehicle tire to obtain vehicle tire local images at different positions.

3. The method of claim 2, wherein The step S2 comprises the following steps: S21, according to the optical characteristics of the cameras, correcting each vehicle tire local image to obtain a corrected image; S22, removing the overlapping part of each corrected image when splicing to obtain a splicing block; S23, splicing each splicing block in turn according to the camera arrangement order to obtain a vehicle tire image.

4. The method of claim 3, wherein The length of the overlapping part of each corrected image when splicing in the step S22 is: wherein, is the length of the image to be removed from each rectified image on both sides, is the number of cameras, is the circumference of the outer circle of the vehicle tire in the image, is the length of the tire in the image taken by each camera, is the imaging coefficient.

5. The method of claim 1, wherein The step S34 comprises the following steps: S341, dividing the standard brightness tire image into multiple tire image blocks; S342, dividing the pre-stored gray vehicle tire image into multiple pre-stored image blocks in the same division manner; S343, calculating the total amount of gray values of each tire image block and pre-stored image block; S345, subtract the total amount of gray values of the tire image block from the total amount of gray values of the pre-stored image block at the same position to obtain a total difference value; S346, select the tire image block with a total difference value exceeding a total threshold value as a damage position.

6. A system for detecting damage to a vehicle tire according to the method of any one of claims 1 to 5, characterized in that, Comprise: a camera, an image splicing unit and a loss identification unit; the camera is used for collecting local images of vehicle tires at different positions; the image splicing unit is used for splicing multiple local images of vehicle tires at different positions to obtain a vehicle tire image; the loss identification unit is used for comparing the vehicle tire image with a pre-stored gray vehicle tire image to obtain a damage position.

Images