Digital Analysis System For The Analysis Of Woven Fabric Appearance Based On Machine Vision

The structural parameters of woven fabrics and the layout of its color yarns have an important influence on the physical and chemical properties of the woven fabric, which determine the applications of woven fabrics. Nowadays, in the field of textile industry, most of the fabric parameters are evaluated by the inspectors using a magnifying glass. This subjective way used to analyze the fabric structure parameters suffers the disadvantages of time-consuming and lab-intensive, meanwhile the results are easily affected by the subjective factors of the inspectors. With the improvement of automation level in textile industry, how to achieve rapid, accurate fabric parameters extraction and analysis has gradually become bottleneck of assembly line production, quality control and reverse engineering. Therefore, to obtain a quick, accurate method for description of woven appearance structure and the parameter analysis is very necessary.In the paper, the background and significance of the topic selected for the thesis are introduced briefly. The research of the recognition of woven fabric parameters based on digital image analysis is summarized. After evaluating several typical analysis and research methods of fabric structure, this paper has proposed a fabric digital analysis method based on machine vision and designed a digital evaluation system at last.In order to improve the detection precision of the system, a fabric dual-side real-time imaging system has been designed and it can simultaneously collect positive and negative image of the corresponding regions for parameter extraction and analysis of samples. A kind of intelligent image measuring method is put forward, which directly uses fabric image projection information and the relations of correspondence pixel points to realize the automatic calculation of weft and warp density.Then, gray-projection method and filtering algorithm are used to locate the weft and warp yarns to realize fabric warp, weft yarn separation and organization point positioning, and the active grid model of the sample fabric can been established; Classifying the types of interlacing points based on the edge intensities has been realized by using recognition algorithm; At the same time, the organization point adjacent information, organization arrangement pattern and color feature information have been used for the correction of the organization classification results and realizes the accurate extraction of tissue type. The color clustering method is also applied to extract color information of the interlacing points and using the interlacing type and color information as the structural parameters to realize the digital encoding and storage of the fabrics; And in a repeat unit of the fabric grid model, the digital files of fabric sample grid model is constructed as the information of the automatic recognition for the classification of the fabric pattern and color layout.In the paper, the method we proposed not only can be used for the analyzing fabric structure parameters automatically and rapidly, and can be used for the reverse reconstruction of weave pattern and color sorting. The experimental results have shown that the proposed method is effective.