基于最小风险贝叶斯决策的印刷品斑点检测方法  

Prints Blobs Detection Method Based on Minimum Risk Bayesian Decision

作　　者：孙冬[1] 刘丹[1] 赵键[2,3] 

机构地区：[1]河南机电高等专科学校计科系 [2]国防科学技术大学电子科学与工程学院ATR重点实验室 [3]中国人民解放军95380部队

出　　处：《计算机测量与控制》2014年第6期1711-1713,共3页Computer Measurement &Control

基　　金：河南教育厅自然科学研究计划项目(2011B520010)

摘　　要：针对印刷品斑点检测缺乏有效检测方法的难题,依据统计决策和贝叶斯分析的基本原理,结合模式识别和机器礼堂相关理论技术,提出了一种基于最小风险贝叶斯决策的印刷品斑点检测方法;以烟标图像为例,实验验证了所提算法的准确性与有效性;该实验表明:所提算法能够有效地检测出烟标中的黑色斑点,分割出来的斑点图像基本上保持了原有的形状;最后,提出了进一步的改进方法。It is hard to detect prints blobs with the present methods, based on the basic principle of statistical decision and Bayesian anal- ysis, combining the theory of pattern recognition and machine vision related technology, this paper proposes a prints blobs detection method which based on minimum Bayesian risk decision method. A case study of the cigarette label is given to verify the accuracy and validity of this algorithm. The results of experiment show that the black blobs of the cigarette label can be detected quickly by the proposed method, and the blob image can keep the original shape. Finally, the further improvement of the method are proposed.

关 键 词：统计决策论 贝叶斯分析 印刷品斑点检测 最小风险贝叶斯决策 

分 类 号：TP274.5[自动化与计算机技术—检测技术与自动化装置]