基于纹理边缘周期性与局部方向性的织物疵点检测  

Fabric Defect Detection Based on Periodicity and Local Directivity of Texture Edge

作　　者：吴宁[1] 管声启[1] 徐帅华 

机构地区：[1]西安工程大学机电工程学院,陕西西安710048

出　　处：《计算机与现代化》2014年第4期16-19,共4页Computer and Modernization

基　　金：陕西省教育厅专项基金资助项目(2013JK1083);西安工程大学博士启动基金资助项目(BS1005)

摘　　要：根据织物纹理图像自身的特点,提出一种基于纹理边缘周期性与局部方向性的织物疵点检测新方法。通过正常纹理边缘的周期性与方向性特征剔除同类有疵点纹理图像的背景纹理信息,突出疵点信息,进而快速有效地检测出无特定方向的织物疵点。经过对大量常见织物疵点图像的检测实验,表明该方法对于纹理边缘清晰、方向一致性较好的织物疵点图像具有较好的检测效果。According to the characteristic of the fabric texture image itself, a new method of fabric defect detection is presented, which based on texture periodicity and local direetivity. Through eliminating the background texture information of similar texture with defect using its features of periodicity and directivity of normal texture edge, the defect information is highlighted, and then the defect information of no particular direction is detected quickly and efficiently. Through detection experiments of a large num- ber of fabric image with defect, this method is proved to have good detection effect, when the fabric image has clear texture edge or good direction consistency.

关 键 词：疵点检测 纹理边缘 纹理周期 局部方向 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]