CN-121982339-A - Plastic bag raw material characteristic extraction method and system based on visual detection

CN121982339ACN 121982339 ACN121982339 ACN 121982339ACN-121982339-A

Abstract

The embodiment of the application provides a plastic bag raw material characteristic extraction method and system based on visual detection, and relates to the technical field of visual detection, wherein the method comprises the steps of obtaining an image to be processed of a plastic film raw material moving at a high speed; performing brightness equalization processing and denoising processing on the image to be processed to obtain a target image, performing feature analysis on the target image to obtain a feature analysis result, wherein the feature analysis comprises enhanced micro-feature analysis or enhanced internal light and shadow behavior pattern analysis, performing local color consistency evaluation on the target image to obtain a color recognition result, and generating a sorting control instruction according to the feature analysis result and the color recognition result. The application can improve the identification precision of an automatic sorting system and ensure the purity and quality of the regenerated plastic particles.

Inventors

ZOU BINGLIN
ZENG LINGLI
WANG LINFENG

Assignees

深圳市万胜环保科技有限公司

Dates

Publication Date: 20260505
Application Date: 20260126

Claims (10)

1. The plastic bag raw material characteristic extraction method based on visual detection is characterized by comprising the following steps of: Acquiring an image to be processed of a plastic film raw material moving at a high speed; Performing brightness equalization processing and denoising processing on the image to be processed to obtain a target image; performing feature analysis on the target image to obtain a feature analysis result, wherein the feature analysis comprises enhanced micro-feature analysis or enhanced internal light and shadow behavior pattern analysis; performing local color consistency evaluation on the target image to obtain a color recognition result; and generating a sorting control instruction according to the characteristic analysis result and the color recognition result.
2. The method according to claim 1, wherein performing brightness equalization and denoising on the image to be processed to obtain a target image comprises: adjusting the overall brightness and contrast of the image to be processed through a self-adaptive histogram equalization algorithm to obtain an initial image; and removing random noise in the initial image by adopting a non-local mean denoising algorithm to the initial image so as to obtain a target image.
3. The method of claim 1, wherein when the feature analysis is an enhanced micro-feature analysis, the performing feature analysis on the target image to obtain a feature analysis result comprises: performing edge morphology analysis on the target image to obtain an analysis result; Detecting the internal shadow anomaly of the target image, and identifying an anomaly region which is obviously different from the standard texture of the single-layer film in the target image; And obtaining a characteristic analysis result according to the analysis result and the abnormal region.
4. A method according to claim 3, wherein the performing edge morphology analysis on the target image to obtain an analysis result comprises: Performing edge morphology analysis on the target image to obtain a contour line of the target image; Calculating the brightness gradient change rate of the edge area of the target image and the width of the gradient change area based on the contour line; And obtaining an analysis result according to the brightness gradient change rate and the width of the gradient change area.
5. The method according to claim 1, wherein the performing the local color consistency evaluation on the target image to obtain a color recognition result includes: converting the target image into an HSV color space image, and dividing the HSV color space image into a plurality of local areas; calculating statistics of hue and saturation characteristics of each of the local regions; acquiring statistics of the overall hue and saturation characteristics of the target image; And comparing the tone and saturation statistics of each local area with the statistics of the integral tone and saturation characteristics of the target image to obtain a color recognition result.
6. The method of claim 1, wherein when the feature analysis is an enhanced internal light and shadow behavior pattern analysis, the performing feature analysis on the target image to obtain a feature analysis result comprises: Extracting local shadow features of the target image to obtain shadow feature vectors; Performing light and shadow behavior pattern recognition and evolution track tracking on the light and shadow feature vector to obtain a behavior pattern; and obtaining a feature analysis result according to the shadow feature vector and the behavior mode.
7. The method of claim 6, wherein the performing local shadow feature extraction on the target image to obtain a shadow feature vector comprises: Convolving the target image to obtain a shadow abnormal response diagram; Carrying out local texture analysis on the shadow abnormal response graph to obtain gray level co-occurrence matrix characteristics; And obtaining a shadow feature vector according to the gray level co-occurrence matrix feature.
8. The method of claim 6, wherein performing shadow behavior pattern recognition and evolution trajectory tracking on the shadow feature vector to obtain a behavior pattern comprises: performing light and shadow behavior pattern recognition and evolution track tracking on the light and shadow feature vector to obtain a feature standard deviation; And judging according to the characteristic standard deviation and the preset standard deviation to obtain a behavior mode.
9. The method of claim 1, wherein generating sort control instructions from the feature analysis results and the color recognition results comprises: Judging whether the image to be processed is a superimposed body or not according to the characteristic analysis result to obtain a judgment result; Performing material reasoning according to the color recognition result to obtain a material type; and generating a sorting control instruction according to the judging result and the material type.
10. A plastic bag raw material characteristic extraction system based on visual detection is characterized by comprising: the acquisition module is used for acquiring images to be processed of the plastic film raw materials moving at a high speed; the preprocessing module is used for carrying out brightness equalization processing and denoising processing on the image to be processed to obtain a target image; The feature analysis module is used for carrying out feature analysis on the target image to obtain a feature analysis result, wherein the feature analysis comprises enhanced micro-feature analysis or enhanced internal light and shadow behavior pattern analysis; the evaluation module is used for carrying out local color consistency evaluation on the target image to obtain a color recognition result; And the generation instruction module is used for generating a sorting control instruction according to the characteristic analysis result and the color recognition result.

Description

Plastic bag raw material characteristic extraction method and system based on visual detection Technical Field The application relates to the technical field of visual detection, in particular to a plastic bag raw material characteristic extraction method and system based on visual detection. Background In the related art, in modern plastic recycling facilities, an automatic sorting system is a key link for improving efficiency and quality of regenerated products. The feature extraction method and the system based on visual detection play a core role, capture images of mixed waste materials on a conveyor belt through an industrial camera, and identify different types of plastic raw materials by utilizing an image processing technology. However, in a practical industrial setting, such visual inspection systems face a number of persistent challenges, especially when dealing with lightweight, flexible plastic film stock. In actual production operation, the crushed and washed plastic film material, especially polyethylene or polypropylene film with soft texture and large surface area, is easy to curl and fold during conveying and vibrating due to own folds and irregular shapes. When a blue high density polyethylene plastic bag chip of originally large area is kneaded into a cluster, it may appear in the camera view as an irregular small patch, most of its surface area and features are obscured by its own folds. When the system analyzes the small color block, the obtained color information is accurate, but the internal structure and the real unfolding form of the folding color block cannot be perceived, so that the color block can be misjudged as unrecognizable impurities due to irregular form or the size and the quality of the color block are underestimated in subsequent characteristic judgment, and the classification decision is deviated. Disclosure of Invention The present application aims to solve at least one of the technical problems existing in the prior art. Therefore, the application provides a plastic bag raw material characteristic extraction method and system based on visual detection, which aim to improve the identification precision of an automatic sorting system and ensure the purity and quality of regenerated plastic particles. In a first aspect, an embodiment of the present application provides a method for extracting characteristics of a plastic bag raw material based on visual detection, including: Acquiring an image to be processed of a plastic film raw material moving at a high speed; Performing brightness equalization processing and denoising processing on the image to be processed to obtain a target image; performing feature analysis on the target image to obtain a feature analysis result, wherein the feature analysis comprises enhanced micro-feature analysis or enhanced internal light and shadow behavior pattern analysis; performing local color consistency evaluation on the target image to obtain a color recognition result; and generating a sorting control instruction according to the characteristic analysis result and the color recognition result. According to some embodiments of the present application, the performing brightness equalization processing and denoising processing on the image to be processed to obtain a target image includes: adjusting the overall brightness and contrast of the image to be processed through a self-adaptive histogram equalization algorithm to obtain an initial image; and removing random noise in the initial image by adopting a non-local mean denoising algorithm to the initial image so as to obtain a target image. According to some embodiments of the application, when the feature analysis is an enhanced micro-feature analysis, the feature analysis is performed on the target image to obtain a feature analysis result, including: performing edge morphology analysis on the target image to obtain an analysis result; Detecting the internal shadow anomaly of the target image, and identifying an anomaly region which is obviously different from the standard texture of the single-layer film in the target image; And obtaining a characteristic analysis result according to the analysis result and the abnormal region. According to some embodiments of the application, the performing edge morphology analysis on the target image to obtain an analysis result includes: Performing edge morphology analysis on the target image to obtain a contour line of the target image; Calculating the brightness gradient change rate of the edge area of the target image and the width of the gradient change area based on the contour line; And obtaining an analysis result according to the brightness gradient change rate and the width of the gradient change area. According to some embodiments of the present application, the performing local color consistency evaluation on the target image to obtain a color recognition result includes: converting the target image into an HSV color space image, and