CN-121186055-B - Artificial leather defect detection method

Abstract

The invention relates to the technical field of machine vision and discloses an artificial leather defect detection method which comprises the steps of extracting texture images of an artificial leather image to be detected, comparing and analyzing the texture images with corresponding preset texture images to obtain a surface defect area, correcting n time-series infrared images of the same artificial leather to be detected based on the texture areas and the surface defect areas, eliminating false positive interlayer defects of the infrared images caused by the surface defect areas, obtaining corrected infrared images, extracting dynamic thermal response characteristics based on the n time-series corrected infrared images, identifying interlayer defect areas and interlayer defect types according to the n corrected infrared images and the dynamic thermal response characteristics, and being suitable for high-quality artificial leather with complex textures on the surface, and the detection accuracy is remarkably improved.

Inventors

• Hong Dongquan

Assignees

• 佛山市三水南基塑胶制品有限公司

Dates

Publication Date

20260508

Application Date

20251011

Claims (3)

1. 1. A method for detecting defects in artificial leather, comprising: Extracting a texture map of an artificial leather image to be detected; Comparing and analyzing the texture map with a corresponding preset texture map to obtain a surface defect area, wherein the preset texture map is a texture map of the same model of the artificial leather product corresponding to the texture map; correcting n time-series infrared images of the same artificial leather to be detected based on a texture area and a surface defect area, eliminating false positive interlayer defects of the infrared images caused by the surface defect area, obtaining corrected infrared images, constructing a first prediction model for predicting first filling temperature of the texture area based on ambient temperature, texture width, texture depth and temperature outside the texture area, setting the temperature of the texture area at the position of the infrared images as the first filling temperature, obtaining corrected infrared images, constructing a second prediction model for predicting second filling temperature of the surface defect area based on the ambient temperature, the width of the surface defect area, the depth of the surface defect area and the temperature outside the texture area, and setting the temperature of the surface defect area at the position of the infrared images as the second filling temperature, wherein the surface defect area depth calculating method comprises the following steps: Drawing a straight line along a direction perpendicular to or penetrating through the surface defect area by using a line contour tool, and sequentially covering the surface of the artificial leather product, the edge of the surface defect area, the bottom of the surface defect area, the other edge of the surface defect area and the surface of the artificial leather product; Generating a gray value-pixel position curve, wherein a valley of the curve corresponds to the bottom of a surface defect area, peaks at two sides correspond to the surface of an artificial leather product, and the pixel distance difference between the valley and the peaks is converted into actual depth, namely the depth of the surface defect area through a preset mapping relation; Based on the n corrected infrared images in time series, extracting dynamic thermal response characteristics, and identifying interlayer defect areas and interlayer defect types according to the n corrected infrared images and the dynamic thermal response characteristics, wherein the dynamic thermal response characteristics comprise heating rate, cooling rate, peak temperature and peak temperature reaching time of each pixel block, and the dynamic thermal response characteristic acquisition method comprises the following steps: drawing a temperature-time curve for each pixel block in the infrared image after n time series correction, and extracting the heating rate, the cooling rate, the peak temperature and the time reaching the peak temperature of each pixel block according to the temperature-time curve; The method for identifying the interlayer defect region and the interlayer defect type according to the dynamic thermal response comprises the following steps: Constructing a data set, wherein the data set comprises a plurality of groups of input features and output features corresponding to the input features, the input features comprise an nth corrected infrared image and dynamic thermal response features, the output features are interlayer defect types and interlayer defect areas, the data set is divided into a training set and a verification set, an improved semantic segmentation model based on U-Net is selected, the training set is used for training the model, model parameters are optimized through the verification set, the trained model is used for segmenting the nth corrected infrared image, and a mask image and defect types of the interlayer defect areas are directly output.
2. 2. The method for detecting the defects of the artificial leather according to claim 1, wherein the defect detection information comprises the positions of interlayer defect regions and the types of interlayer defects, the positions of surface region defects or whether the same region coexists with the interlayer defect regions and the surface defect regions.
3. 3. The method for detecting the defects of the artificial leather according to claim 1, wherein the method for acquiring the surface defect areas comprises the following steps: Step 1, adding a texture map into a virtual frame, placing the virtual frame in a plane coordinate system, wherein the virtual frame is a frame formed by edges of an artificial leather product, and the position of the texture map in the virtual frame is consistent with the position of the texture map on the surface of the artificial leather product; step 2, replacing the texture map in the step 1 with a preset texture map, and superposing the preset texture map into the virtual frame in the same manner; And 3, the texture of the preset texture map and the texture map is composed of a plurality of adjacent pixel blocks, each pixel block has corresponding coordinates in the virtual frame, if two identical coordinates are detected in the virtual frame, the preset texture map and the texture of the texture map are indicated to correspond, if only one pixel coordinate is detected in the virtual frame, the pixel block coordinates are marked as defective pixel block coordinates, and the adjacent defective pixel block coordinates form a surface defect area.

Description

Artificial leather defect detection method Technical Field The invention relates to the technical field of machine vision, in particular to a method for detecting defects of artificial leather. Background The defects of the artificial leather as a multi-layer composite material are mainly divided into surface defects and interlayer defects, wherein the interlayer defects are hidden in the material, so that the influence on the durability of the product is larger, but the product is more difficult to detect. The traditional artificial leather defect detection is based on a plane vision technology, such as a common camera device, only can identify surface visible defects and cannot capture interlayer defects, even if part of schemes introduce infrared thermal imaging technology to try to distinguish the interlayer defects through thermal response differences, such as abnormal heating/cooling rate caused by poor bubble heat conduction, core bottlenecks still face, and when the artificial leather surface is provided with complicated textures such as imitation wood grains, imitation leather grains and the like, the geometric structures of the textures, such as bulges or depressions, change local heat conduction characteristics, so that the heating and cooling rates of the artificial leather are highly similar to the thermal response characteristics of the interlayer defects, and serious interference is formed. The traditional method cannot effectively distinguish the thermal response abnormality caused by textures from the thermal response abnormality of real interlayer defects, false positive misjudgment (misjudgment of textures as interlayer defects) or missed judgment (masking of interlayer defects by textures) easily occur, and finally, the high-quality artificial leather with complex textures on the surface is low in defect detection accuracy, and the requirement of industrial production on detection accuracy is difficult to meet. Therefore, a method for detecting defects of artificial leather is provided to solve the difficulties existing in the prior art, which is a problem to be solved by those skilled in the art. Disclosure of Invention The invention aims to provide an artificial leather defect detection method for solving the defects in the background technology. In order to achieve the above object, the present invention provides the following technical solutions: A method for detecting defects of artificial leather, comprising: Extracting a texture map of an artificial leather image to be detected; Comparing and analyzing the texture map with a corresponding preset texture map to obtain a surface defect area, wherein the preset texture map is a texture map of the same model of the artificial leather product corresponding to the texture map; Correcting n time series infrared images of the same artificial leather to be detected based on the texture area and the surface defect area, eliminating false positive interlayer defects of the infrared images caused by the surface defect area, and obtaining corrected infrared images; based on the n corrected infrared images in time series, extracting dynamic thermal response characteristics, and identifying interlayer defect areas and interlayer defect types according to the n corrected infrared images and the dynamic thermal response characteristics; And obtaining defect detection information of the artificial leather image to be detected based on the surface defect area position, the interlayer defect area position and the defect type of the artificial leather image. Further, the defect detection information includes an interlayer defect region position, an interlayer defect type thereof, a surface region defect position, or whether the same region coexists with the interlayer defect region and the surface defect region. Further, the surface defect area obtaining method comprises the following steps: Step 1, adding a texture map into a virtual frame, placing the virtual frame in a plane coordinate system, wherein the virtual frame is a frame formed by edges of an artificial leather product, and the position of the texture map in the virtual frame is consistent with the position of the texture map on the surface of the artificial leather product; Step 2, replacing the texture map in step 1 with a preset texture map, and performing the same manner Superposing a preset texture map into the virtual frame; And 3, the texture of the preset texture map and the texture map is composed of a plurality of adjacent pixel blocks, each pixel block has corresponding coordinates in the virtual frame, if two identical coordinates are detected in the virtual frame, the preset texture map and the texture of the texture map are indicated to correspond, if only one pixel coordinate is detected in the virtual frame, the pixel block coordinates are marked as defective pixel block coordinates, and the adjacent defective pixel block coordinates form a surface defect area. Further, a f