基于计算机视觉的皮革检测技术研究  

Leather Detection Technology Based on Computer Vision

作　　者：张娜[1] ZHANG Na(Shaanxi Electronic Information Institute,Xi'an 710500,China)

机构地区：[1]陕西电子信息职业技术学院,陕西西安710500

出　　处：《中国皮革》2022年第12期43-47,共5页China Leather

基　　金：陕西省教育科学“十三五”规划2018年度课题(SGH18H552)。

摘　　要：皮革表面难免存在破损、划痕、褶皱等瑕疵,为提高产品合格率,对皮革进行质量检测显得尤为重要。长期以来,皮革检测普遍由人工完成,然而人工检测存在效率低、成本高且误检率高等不足之处。鉴于皮革表面纹理、光泽、色彩都具有鲜明的特点,将计算机视觉技术用于皮革正反面识别及缺陷检测方法的研究中。基于图像处理系统组成框架,借助模式识别中的人工神经网络设计分类器,经试验表明,人工神经网络三层结构在皮革正反面识别的应用中具有较高的准确率。本文将统计学中的共生矩阵思想引入皮革图像纹理特征提取的研究中,并提出一种基于模糊算法的C-均值聚类方法,可精准高效识别皮革表面缺陷。It is inevitable that there are damages, scratches, wrinkles and other defects on the leather surface. In order to improve the grid rate of products, it is particularly important to detect the quality of leather. For a long time, leather detection has been generally completed manually. However, manual detection has some disadvantages, such as low efficiency, high cost and high false detection rate. In view of the distinctive characteristics of leather surface texture, luster and color, computer vision technology were applied to the research of leather front and back recognition and defect detection methods. Based on the framework of image processing system, the classifier was designed with the help of artificial neural network in pattern recognition. The experimental results show that the three-layer structure of artificial neural network has a high accuracy in the application of leather front and back recognition. The idea of co-occurrence matrix in statistics was introduced into the research of leather image texture feature extraction, and a C-means clustering method based on fuzzy algorithm was proposed, which can accurately and efficiently identify leather surface defects.

关 键 词：计算机视觉 人工神经网络 C-均值聚类 缺陷检测 

分 类 号：TS57[轻工技术与工程—皮革化学与工程]