一种基于连通域标记的纸病检测算法  

Algorithm of Paper Defect Detection Based on Connected-component Labeling

作　　者：赵晓 何立风 姚斌 高启航 杨云 ZHAO Xiao;HE Li-feng;YAO Bin;GAO Qi-hang;YANG Yun(Artificial lntdligence Institute,College of Electrical and Information Engineering,3haanxi University of Science ＆ Technology,Xi＇an,3haanxi Province,710021;Faculty of Information Science and Technology,Aichi Prefectural University,Aichi,Japan,4801198)

机构地区：[1]陕西科技大学电气与信息工程学院人工智能研究所,陕西西安710021 [2]爱知县立大学信息科学学院,日本爱知4801198

出　　处：《中国造纸学报》2018年第2期51-56,共6页Transactions of China Pulp and Paper

基　　金：国家自然基金项目(61601271;61471227;61603234);陕西省科技项目(2016SF-444);陕西省教育厅科研项目(16JK1087)

摘　　要：提出了一种基于段的连通域标记处理算法,同时对纸病区域进行连通域标记和形状特征值提取,旨在提高纸病检测的准确率和效率。该算法利用纸病区域为简单连通图像的特点,采用段技术实现了纸病区域的标记处理,探讨了在标记处理同时快速统计与形状特征值计算有关的中间参数的方法,利用标记结果及形状特征值实现了纸病的快速检测。该算法优化了标记处理与形状特征值提取的过程,减少了纸病图像的扫描次数。结果表明,该算法达到了准确、快速的纸病检测效果,且易于扩展到实际的纸病检测系统中。in order to enhance the accuracy and efficiency of paper defect detection, this paper presented a run-based conneeted-componenl labeling algorithm, which could obtain the features of paper defect regions during the connected component was labeled. According to the regions of paper defects are simple connected image, we used a run-based connected-component labeling to record the intermediate results related to the image shape feature extraction during the labeling processing, and combined the labeled results and shape feature values to realize the paper defect detection rapidly. This algorithm could reduce the times of scanning images because it optimized the process of connectedcomponent labeling and shape feature extraction. The experimental results demonstrated that the proposed ulgorithm could achieve an accurate and efficient detection result for general paper defects, and easily be used in the actual paper defect detection system.

关 键 词：连通域标记 段标记 形状特征提取 纸病检测 

分 类 号：TP39[自动化与计算机技术—计算机应用技术]