CN-121994810-A - Visual flaw detection method, system and equipment for car cover base film

CN121994810ACN 121994810 ACN121994810 ACN 121994810ACN-121994810-A

Abstract

The application relates to the field of computer vision, in particular to a flaw vision detection method, a flaw vision detection system and flaw vision detection equipment for a vehicle-mounted basement membrane, wherein the method comprises the steps of determining an abnormal coefficient of an image according to a gray value and an area of the abnormal region in each frame of the image so as to obtain an abnormal coefficient sequence; the method comprises the steps of analyzing extreme values in an air pressure sequence to obtain an oscillation frequency sequence and a target air pressure sequence, determining a high-density area corresponding to each abnormal coefficient based on the distribution density and the space weight of each data point in a two-dimensional section corresponding to each abnormal coefficient, extracting target data points of the high-density area, performing curve fitting on the real-time oscillation frequency sequence and the real-time target air pressure sequence to obtain an operation trend curve, performing curve fitting on each corresponding target data point according to the sequence from the small abnormal coefficient to the large abnormal coefficient to obtain a target trend curve, and outputting a corresponding detection result according to the similarity between the target trend curve and the operation trend curve. The application can improve the defect recognition accuracy of the base film.

Inventors

Zhan hedong
Liang Laigen
LUO DAN
JIANG GUOWEI
YANG HU

Assignees

浙江安布雷拉新材料制造有限公司

Dates

Publication Date: 20260508
Application Date: 20260409

Claims (10)

1. A visual inspection method for flaws in a vehicle-coating base film, the method comprising: continuously acquiring multi-frame images in the transmission process of the base film, and determining an abnormal coefficient of the images according to the gray value and the area of an abnormal region in each frame of images to obtain an abnormal coefficient sequence; The method comprises the steps of analyzing extreme values in an air pressure sequence to obtain a corresponding oscillation frequency sequence and a target air pressure sequence, wherein the air pressure sequence is obtained by continuously obtaining an air pressure value of a pressure stabilizing tank in the conveying process of a base film, the oscillation frequency sequence is obtained by calculating the reciprocal of a difference value of a timestamp corresponding to an adjacent extreme value, and the target air pressure sequence is obtained by calculating the average value of the adjacent extreme values; Determining a high-density region corresponding to each abnormal coefficient based on the distribution density and the space weight of each data point in the two-dimensional section corresponding to each abnormal coefficient in a three-dimensional space, and extracting a target data point of the high-density region; Performing curve fitting on the real-time oscillation frequency sequence and the real-time target air pressure sequence to obtain an operation trend curve, and performing curve fitting on corresponding target data points according to the sequence from small to large of the abnormal coefficients to obtain a target trend curve; and outputting a corresponding detection result according to the similarity between the target trend curve and the running trend curve.
2. The method for detecting the defect vision of the car cover base film according to claim 1, wherein the steps of continuously collecting a plurality of frames of images in the process of conveying the base film, determining the abnormal coefficient of the image according to the gray value and the area of the abnormal region in each frame of images to obtain an abnormal coefficient sequence comprise the following steps: Continuously acquiring multi-frame images in the process of conveying the base film, and carrying out threshold segmentation on each image based on a preset threshold value to obtain an abnormal region of the image; determining a thickness coefficient of each abnormal region according to the difference value between the average pixel value of each abnormal region and the average pixel value of the image to which the abnormal region belongs; carrying out normalization processing on each abnormal region according to the area of the largest abnormal region screened out from the multi-frame image to obtain the proportion weight of the abnormal region; obtaining an abnormality index of each abnormal region according to the product of the proportional weight of the abnormal region and the thickness coefficient of the abnormal region; Calculating the average value of the abnormality indexes of a plurality of abnormality areas in each frame of image to obtain the abnormality coefficient of the image; and sequencing the abnormal coefficients of the multi-frame images according to the time sequence to obtain an abnormal coefficient sequence.
3. The method for visually inspecting defects of a vehicle-mounted substrate film according to claim 1, wherein the analyzing the extremum in the air pressure sequence to obtain the corresponding oscillation frequency sequence and target air pressure sequence comprises: Continuously acquiring the air pressure value of the pressure stabilizing tank in the conveying process of the base film to obtain an air pressure sequence; Calculating a gradient for the air pressure sequence, determining a plurality of extreme values of the air pressure sequence according to the sign change points of the gradient, and sequencing the extreme values according to a time sequence to obtain an extreme value sequence, wherein the extreme values comprise peaks and valleys; Calculating the reciprocal of the difference value of the time stamp corresponding to each adjacent extremum in the extremum sequence to obtain the oscillation frequency so as to obtain an oscillation frequency sequence; and calculating the average value of each adjacent peak value and valley value in the extremum sequence to obtain the target air pressure so as to obtain the target air pressure sequence.
4. The method for visual inspection of defects in a vehicular coating film according to claim 1, wherein the determining a high-density region corresponding to each anomaly coefficient based on a distribution density and a spatial weight of each data point in a two-dimensional cross section corresponding to the anomaly coefficient in a three-dimensional space, and extracting a target data point of the high-density region comprises: Constructing a three-dimensional space according to the abnormal coefficient sequence, the oscillation frequency sequence and the target air pressure sequence three-dimensional point set based on time alignment; Aiming at a two-dimensional section corresponding to each abnormal coefficient, constructing a circular area in a neighborhood by taking each data point in the two-dimensional section as a central point, acquiring the number of the data points contained in the circular area under different radiuses, and calculating the local density coefficient of the data point under the radius based on the inverse relation between the number of the data points and the radius; and determining a high-density region corresponding to each abnormal coefficient according to the local density coefficient and the spatial weight of each data point in the two-dimensional section corresponding to each abnormal coefficient, and extracting a target data point of the high-density region.
5. The method for visual inspection of defects in a vehicular coating film according to claim 4, wherein determining a high-density region corresponding to each anomaly coefficient according to the local density coefficient and the spatial weight of each data point in the two-dimensional cross section corresponding to the anomaly coefficient, and extracting a target data point of the high-density region comprises: aiming at the two-dimensional section corresponding to each abnormal coefficient, calculating the weight coefficient of the abnormal coefficient according to the total number of a plurality of data points in the two-dimensional section, the maximum value of the oscillation frequency in the two-dimensional section and the maximum value of the target air pressure in the two-dimensional section; and determining a high-density region corresponding to the abnormal coefficient according to the local density coefficient of each data point in the two-dimensional section corresponding to each abnormal coefficient and the weight coefficient of the abnormal coefficient, and extracting a target data point of the high-density region.
6. The method for visual inspection of defects in a vehicle-cover base film according to claim 5, wherein determining a high-density region corresponding to each anomaly coefficient according to a local density coefficient of each data point in a two-dimensional cross section corresponding to the anomaly coefficient and a weight coefficient of the anomaly coefficient, and extracting a target data point of the high-density region comprises: Aiming at the two-dimensional section corresponding to each abnormal coefficient, the product of the local density coefficient of each data point under different radiuses and the weight coefficient of the abnormal coefficient is used for determining the density coefficient of the data point under the radius; And selecting the maximum density coefficient from the two-dimensional sections corresponding to each abnormal coefficient, determining a circular area formed by the data point corresponding to the density coefficient and the radius as a high-density area, and determining the data point as a target data point of the high-density area.
7. The method for detecting the flaw vision of the car cover basement membrane according to claim 1, wherein the curve fitting is performed on the real-time oscillation frequency sequence and the real-time target air pressure sequence to obtain an operation trend curve, and the curve fitting is performed on corresponding target data points according to the order of the anomaly coefficient from small to large to obtain the target trend curve, and the method comprises the following steps: According to the sequence from small to large of the abnormal coefficients, arranging all target data points corresponding to the abnormal coefficients in a preset abnormal interval, establishing a two-dimensional coordinate system by taking the oscillation frequency corresponding to each target data point as a horizontal axis and the target air pressure as a vertical axis, and fitting all the target data points according to curve fitting to obtain a target trend curve; And aligning the real-time oscillation frequency sequence and the real-time target air pressure sequence which are synchronously acquired in the current production process according to the time stamp, mapping the real-time oscillation frequency sequence and the real-time target air pressure sequence into the two-dimensional coordinate system, and fitting by adopting curve fitting to obtain an operation trend curve.
8. The visual inspection method of defects of a car cover base film according to claim 1, wherein the outputting the corresponding inspection result according to the similarity between the target trend curve and the running trend curve comprises: according to the DTW algorithm, calculating the similarity between the target trend curve and the running trend curve; Outputting a first detection result in response to the similarity being greater than a first preset value, wherein the first detection result is used for prompting that the equipment has a blocking risk; And outputting a second detection result in response to the similarity being greater than the first preset value and less than the second preset value, wherein the second detection result is used for prompting that the equipment is blocked, and the second preset value is greater than the first preset value.
9. A visual inspection system for flaws in a vehicle skin base film, the system comprising: The acquisition module is used for continuously acquiring multi-frame images in the transmission process of the base film, and determining an abnormal coefficient of the image according to the gray value and the area of an abnormal region in each frame of image to obtain an abnormal coefficient sequence; the system comprises an analysis module, a base film, a pressure stabilizing tank, a target pressure sequence, a vibration frequency sequence and a time stamp, wherein the analysis module is used for analyzing extreme values in the pressure sequence to obtain a corresponding vibration frequency sequence and the target pressure sequence; The system comprises a determining module, a three-dimensional space constructing module, a data acquisition module and a data acquisition module, wherein the determining module is used for determining a high-density area corresponding to each abnormal coefficient based on the distribution density and the space weight of each data point in a two-dimensional section corresponding to each abnormal coefficient and extracting a target data point of the high-density area; The fitting module is used for performing curve fitting on the real-time oscillation frequency sequence and the real-time target air pressure sequence to obtain an operation trend curve, and performing curve fitting on corresponding target data points according to the sequence from small to large of the abnormal coefficient to obtain a target trend curve; And the output module is used for outputting a corresponding detection result according to the similarity between the target trend curve and the running trend curve.
10. An electronic device comprising a memory, a processor and a computer program stored on the memory for execution on the processor, wherein the processor implements the method of any one of claims 1 to 8 when executing the computer program.

Description

Visual flaw detection method, system and equipment for car cover base film Technical Field The application relates to the technical field of computer vision, in particular to a flaw visual detection method, system and equipment for a car cover base film. Background The TPU (Thermoplastic Polyurethane, thermoplastic polyurethane elastomer) base film is the core layer of the invisible car cover and is made of thermoplastic polyurethane. It has excellent elasticity, wear resistance, yellowing resistance and self-repairing performance, and is mainly produced by a tape casting process. In the production process of the car cover base film, the air knife edge is responsible for sending stable and uniform high-speed air flow, the base film material in a high-temperature and viscous flow state is pressed on the surface of the cooling roller, so that the rapid and uniform cooling and shaping of the base film material are ensured, the air knife edge is easy to block due to the accumulation of impurities in the working process, the air flow speed and the pressure at a blocking point are reduced, the base film material in a corresponding area cannot be tightly attached to the cooling roller, the base film material is unevenly cooled at different transverse positions, so that transverse stripes appear, the blocking of the air knife edge cannot be self-healed, unless the blocking degree is manually cleared, the blocking degree is only more and more serious, more and more serious flaws appear in the produced base film, so that the flaws are required to be identified at the initial stage of blocking, the flaws are improved while the loss is avoided, the flaws caused to the base film at the initial stage of blocking of the air knife edge are very slight, no obvious image characteristic is displayed, and the flaws are difficult to identify by a line mode. Disclosure of Invention In order to solve the technical problems, the application aims to provide a visual flaw detection method, a visual flaw detection system and visual flaw detection equipment for a vehicle coating base film, and the adopted technical scheme is as follows: In a first aspect, a method for visually inspecting flaws in a vehicle-mounted base film is provided, the method comprising: continuously acquiring multi-frame images in the transmission process of the base film, and determining an abnormal coefficient of the images according to the gray value and the area of an abnormal region in each frame of images to obtain an abnormal coefficient sequence; The method comprises the steps of analyzing extreme values in an air pressure sequence to obtain a corresponding oscillation frequency sequence and a target air pressure sequence, wherein the air pressure sequence is obtained by continuously obtaining an air pressure value of a pressure stabilizing tank in the conveying process of a base film, the oscillation frequency sequence is obtained by calculating the reciprocal of a difference value of a timestamp corresponding to an adjacent extreme value, and the target air pressure sequence is obtained by calculating the average value of the adjacent extreme values; Determining a high-density region corresponding to each abnormal coefficient based on the distribution density and the space weight of each data point in the two-dimensional section corresponding to each abnormal coefficient in a three-dimensional space, and extracting a target data point of the high-density region; Performing curve fitting on the real-time oscillation frequency sequence and the real-time target air pressure sequence to obtain an operation trend curve, and performing curve fitting on corresponding target data points according to the sequence from small to large of the abnormal coefficients to obtain a target trend curve; and outputting a corresponding detection result according to the similarity between the target trend curve and the running trend curve. Optionally, continuously acquiring a plurality of frames of images in the process of conveying the base film, determining an abnormal coefficient of the image according to the gray value and the area of an abnormal region in each frame of images to obtain an abnormal coefficient sequence, including: Continuously acquiring multi-frame images in the process of conveying the base film, and carrying out threshold segmentation on each image based on a preset threshold value to obtain an abnormal region of the image; determining a thickness coefficient of each abnormal region according to the difference value between the average pixel value of each abnormal region and the average pixel value of the image to which the abnormal region belongs; carrying out normalization processing on each abnormal region according to the area of the largest abnormal region screened out from the multi-frame image to obtain the proportion weight of the abnormal region; obtaining an abnormality index of each abnormal region according to the product of the proportiona