一种新的粘连字符图像分割方法  

A new segmentation method for touching character images

作　　者：孔月萍[1] 郭世雄[1] 梁韶军[2] 

机构地区：[1]西安建筑科技大学信息与控制工程学院,陕西西安710055 [2]西安市中心医院麻醉科,陕西西安710003

出　　处：《电子技术应用》2009年第7期136-138,共3页Application of Electronic Technique

基　　金：西安市卫生局科技基金项目(07-7);陕西省教育厅专项计划(04JK244)资助

摘　　要：针对监控画面采样图像中数字的自动识别问题,提出一种新的粘连字符图像分割方法。该方法以预处理后二值图像的连通状况来判定字符粘连的存在,并对粘连字符图像采用上下轮廓极值法确定候选粘连分割点,以双向最短路径确定合适的图像分割线路。仿真实验表明,该方法能有效解决粘连字符图像的分割问题。A new segmentation algorithm is proposed for touching character images for the automatic recognition of characters on monitor screen, By using connected conditions of the image after preprocessing, touching character can be judged, and the cut points candidate of the touching character are located by using the extremum of character contours. Finally the correct cut path is defined by the bidirectional shortest-path. Experimental results show that this method is effective.

关 键 词：字符分割 连通状况 粘连字符 轮廓 最短路径 

分 类 号：TP751[自动化与计算机技术—检测技术与自动化装置]