基于统计阈值法的印品缺陷检测  

Printing Defects Inspection Based on the Statistical Threshold Method

作　　者：官燕燕[1] 刘昕[1] 

机构地区：[1]西安理工大学印刷包装工程学院,陕西西安710048

出　　处：《西安理工大学学报》2007年第4期410-413,共4页Journal of Xi'an University of Technology

基　　金：陕西省自然科学研究计划项目(2004E229)

摘　　要：针对印刷品的缺陷检测问题,在传统模板匹配法的基础上,根据统计过程控制原理,提出利用统计阈值法代替传统的固定阈值法,以自动确定图像上各个像素点的灰度阈值,解决了传统固定阈值法在边缘纹理区域容易产生轮廓误检和漏检的问题。实验结果表明,改进算法相对于传统算法在检测精度和检测速度上都有明显提高。With an aim at printing defects inspection, based on the conventional pattern matching algorithm and in terms of control theory in the statistical process, this paper suggests a statistical threshold method replace the conventional threshold method so as to automatically determine the gray threshold of each pixel in the image,whereby the problems of outline falso inspection and o- mission inspection easily caused by the conventional threshold method in the edge grain zone are solved. The experiment results indicate that the improved algorithm relative to the conventional algorithm is obviously raised in the detection accuracy and velocity.

关 键 词：印品缺陷检测 模板匹配 统计阈值法 灰度阈值 

分 类 号：TS807[轻工技术与工程]