Recognition Algorithms For Woven Fabric Based On Digital Image Processing

The traditional recognition of woven fabric mainly is done manually.This method of manual recognition has obviously low efficiency and dependence on the manual experience,also it may cause damage on the fabric and other issues.In the situation of the rapid development of the textile industry,manual recognition has been far from being able to meet the urgent demand of textile enterprises to reduce labor costs and improve production efficiency.It is important to improve the competitiveness of textile enterprises by studying the recognition algorithms for woven fabric based on digital image processing technology.The main content of this paper is to study the recognition algorithms for woven fabric and the main purpose is to get the fabric pattern grid from the captured fabric image by digital image processing technology.The design of the algorithm mainly includes the following steps:the preprocessing of fabric image,the segmentation of fabric points,the feature extraction and classification of fabric points,as well as the generation and correction of pattern grid.In this paper,a variety of treatment processes are designed in different steps of the recognition of woven fabric.Firstly,Gaussian filter is used to suppress the image noise,and the color fabric image is grayscale.Secondly,the segmentation process of fabric points is studied in detail.The method of gray level gradient integral segmentation based on spatial domain and the method of segmentation based on fabric binary image are respectively applied to the fabric images under different conditions.In addition,according to the color of warp and weft yarns in the fabric samples,the method of generating fabric binary image based on clustering and the method of generating fabric binary image based on Tamura texture feature are proposed to generate fabric binary image.Then,in the process of fabric feature extraction and classification,according to the difference between the width of fabric points and the direction of texture in the fabric points,the method of division based on fabric point width and the method of texture extraction based on Gabor filter are proposed.And the acquisition of fabric width and the selection of the Gabor filter parameters are introduced in detail.Finally,.the fabric points mark image is obtained after fabric points classification.Finally,according to the fabric points mark image,the fabric pattern matrix is generated,and then the standard fabric pattern matrix library is used to correct the fabric pattern matrix.Finally,the pattern grid of fabric image is obtained.The whole set of algorithms studied in this paper can realize the recognition of woven fabric well and improve the efficiency and accuracy of the manual detection,which is of great theoretical significance.