计算机图像处理技术在大面积皮革表面缺陷检测中的应用研究  

Application of Computer Image Processing Technology in Leather Surface Inspection

作　　者：王潇[1] WANG Xiao(Xi'an Fanyi University,Xi'an 710105,China)

机构地区：[1]西安翻译学院,陕西西安710105

出　　处：《中国皮革》2022年第9期53-56,60,共5页China Leather

基　　金：陕西省教育科学“十三五”规划2020年课题(SGH20Y1497);西安翻译学院项目(T1901)。

摘　　要：皮革表面的划痕、疵点、破洞等缺陷不仅影响皮革产品的美观,更重要的是降低了高端皮革制品的品质。由于人工检测大面积皮革表面缺陷的难度较大,因此可采用计算机图像处理技术来提高皮革表面检测效率。首先,进行皮革图像的预处理,基于改进LOG算子对皮革图像进行降噪和锐化,处理后的皮革表面图像细节更明显;其次,基于Gabor小波变换提取皮革表面缺陷区域,为降低运算量,进行图像降维然后分割出缺陷区域,并利用灰度共生矩阵提取皮革表面缺陷特征;最后,采用基于SVM的缺陷分类算法,对皮革表面缺陷的种类进行判别和分类。经试验验证:皮革表面缺陷的分类精度在88.5%左右。The scratches,defects,holes and other defects on the leather surface not only affect the beauty of leather products,but also reduce the quality of high-end leather products.Because it is difficult to detect large-area leather surface defects manually,computer image processing technology can be used to improve the efficiency of leather surface detection.Firstly,the leather image is preprocessed.Based on the improved LOG operator,the leather image is denoised and sharpened,and the details of the processed leather surface image are more obvious.Secondly,the defect region of leather surface is extracted based on Gabor wavelet transform.In order to reduce the amount of computation,the image dimension is reduced,and then the defect region is segmented,and the defect feature of leather surface is extracted by gray level co-occurrence matrix.Finally,the defect classification algorithm based on SVM is used to distinguish and classify the types of leather surface defects.The test shows that the classification accuracy of leather surface defects is about 88.5%.

关 键 词：计算机图像处理技术 大面积皮革表面 缺陷检测 特征提取 缺陷分类 

分 类 号：TS56[轻工技术与工程—皮革化学与工程]