Research On Fabric Defect Detection And Classification Algorithm Based On LTP Gray Level Co-occurrence Matrix And SVM

The problem of fabric quality is very important for textile enterprises.Howto quickly and effectively detect and identify defects in fabrics has become an urgent problem that the textile industry needs to optimize and solve.Due tothe high cost of fabric inspection systems available abroad,most domestic textile companies still use manual cloth inspection methods.Research by various parties has shown that this method is not only inefficient,but also the detection rate of fabric defects and the subjective cognition of workers.The resulting restraint has a great relationship,so it is very meaningful to study the automatic detection and classification methods of fabric defects.This paper mainly studies the detection and classification algorithm of fabric defects,and designs an automatic detection and classification system of fabric defects according to the researched algorithm.Thesis research work mainly includes the following aspects:First of all,pre-process the collected fabric defect images.Due to the complexity and diversity of the texture color of the fabric and the different lighting environments in which the image is collected,it will bring a lot of interference to the detection of fabric defects.To prevent these reasons,the image feature detection The accuracy of the image is affected,and the image is processed with grayscale and the image of the fabric defect is enhanced and denoised by histogram equalization and homomorphic filtering.Secondly,in terms of fabric defect detection and extraction,based on the LBP operator and its optimization model,the LTP operator is improved in rotation unchanged and center symmetry,and combined with the characteristic parameters of the gray level co-occurrence matrix for fabric defects Detection and feature extraction effectively improve the detection rate,reduce the amount of calculation and shorten the detection time.Through the image segmentation andthreshold segmentation,the area of the fabric defect can be accurately segmented.The experiment proves that the maximum inter-class variance method(Ostu)is the most clear for the hole defect image segmentation,and the information of the defect is relatively complete.Non-defective pixel processing is also better.The segmented image is combined with closed operation and open operation for morphological processing,and connected domain analysis is used to locate and mark fabric defects.In terms of fabric defect classification,the SVM multi-classifier and kernelfunction are selected through experiments,and the Gaussian kernel function and1-v-r SVMs multi-classifier are used to identify the fabric defect features extracted by the detection algorithm in this paper.Finally,a fabric defect detection and classification system based on LTP gray level co-occurrence matrix and SVM is designed in combination with Matlab software.The algorithm studied in this paper is further analyzed through experiments to verify the effectiveness of the detection mark positioning and classification algorithm in this paper.