基于小波分解和奇异值分解的织物疵点检测  

Fabric Defect Detection Based on Wavelet Decomposition and Singular Value Decomposition

作　　者：李鹏飞[1] 姜萌[1] 景军锋[1] 张蕾[1] 张宏伟[1] 

机构地区：[1]西安工程大学,陕西西安710048

出　　处：《棉纺织技术》2015年第6期41-44,52,共5页Cotton Textile Technology

基　　金：国家自然科学基金(№61301276);西安工程大学学科建设经费资助基金(№107090811)

摘　　要：探讨基于小波分解和奇异值分解的织物疵点检测效果。运用三种算法进行了织物疵点检测。采用基本奇异值分解法进行机织物疵点检测,检测结果受噪声影响较大;采用Haar小波对待测图像进行除噪,边缘检测性较弱。利用Gabor小波良好的图像边缘敏感度、方向和尺度选择性,进行图像滤波,再进行SVD分解。结果表明:Gabor小波和SVD的融合算法可以对图像多种方向多种尺度进行调节,检测效果较好。认为:Gabor小波和奇异值分解相融合算法可应用于机织物疵点检测。Fabric defect detection effect based on wavelet decomposition and singular value decomposition was discussed. Three kinds of woven fabric defect detection calculating methods were adopted. When basic singular value decomposition method was used for woven fabric defect detection, test results were greatly affected by noise. When Haar wavelet was used on images need to be tested to remove noise,edge detection was weaker. Uti- lizing better image edge sensitiveness and direction dimension selectivity of Gabor wavelet, image filtering was proceeded and then SVD decomposition was implemented, The results show that fusion algorithm of Gabor wavelet and SVD can adjust multiple direction and multiple dimension of images, the detection effect is better. It is considered that fusion algorithm of Gabor wavelet and singular value decomposition can be used in woven fabric defect detection.

关 键 词：织物疵点检测 奇异值分解 HAAR小波 GABOR小波 阈值处理 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]