一种基于边缘形状特征的字符缺陷检测算法  

CHARACTER DEFECT DETECTION ALGORITHM BASED ON EDGE SHAPE FEATURES

作　　者：孙洋 孙林[1] 夹尚丰 余振军 Sun Yang;Sun Lin;Jia Shangfeng;Yu Zhenjun(Geomatics College,Shandong University of Science and Technology,Qingdao 266590,Shandong,China;Qingdao Xingke Ruisheng Information Technology Co.,Ltd.,Qingdao 266590,Shandong,China)

机构地区：[1]山东科技大学测绘科学与工程学院,山东青岛266590 [2]青岛星科瑞升信息科技有限公司,山东青岛266590

出　　处：《计算机应用与软件》2023年第9期177-183,共7页Computer Applications and Software

基　　金：国家自然科学基金项目(41771408);山东省自然科学基金项目(ZR201702210379)。

摘　　要：针对传统的缺陷检测方法在复杂光照条件下对印刷品缺陷检测精度不高的问题,提出一种基于边缘形状特征的字符缺陷检测算法。该算法通过Sobel算子滤波后得到的梯度数据为基础,根据边缘连通域长度,确定最优的边缘分割阈值。获得边缘轮廓点的梯度方向和坐标作为匹配信息,根据轮廓点匹配度,使用粗筛选和精筛选相结合的方式来寻找缺陷点,基于缺陷点填充寻找连通域的方式,标记字符缺陷区域。为了使匹配算法满足实时性要求,采用图像金字塔算法和通过阈值判断提前结束的搜索策略。实验表明该算法在复杂光照条件下能够快速准确定位印刷品缺陷。Aiming at the problem that traditional defect detection methods do not have high accuracy in the detection of print defects under complex lighting conditions,this paper proposes a character defect detection algorithm based on edge shape features.Based on the gradient data obtained by Sobel operator filtering,this algorithm determined the optimal edge segmentation threshold based on the length of the edge connected domain.The gradient direction and coordinates of the edge contour points were obtained as matching information.According to the degree of contour point matching,a combination of coarse screening and fine screening was used to find defect points,and the method of filling connected regions based on defect points was used to mark the defect areas of characters.In order to make the matching algorithm meet the real-time requirements,an image pyramid algorithm and a search strategy that judged to end prematurely by threshold were adopted.Experiments show that the algorithm can quickly and accurately locate print defects under complex lighting conditions.

关 键 词：复杂光照 边缘分割 Two-Pass算法 缺陷检测 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]