基于边缘的字符串定位算法  

Edge-based algorithm for text string location

作　　者：李闯[1] 丁晓青[1] 吴佑寿[1] 

机构地区：[1]清华大学电子工程系智能技术与系统国家重点实验室,北京100084

出　　处：《清华大学学报（自然科学版）》2005年第10期1335-1338,共4页Journal of Tsinghua University(Science and Technology)

基　　金：国家自然科学基金资助项目(60241005)

摘　　要：为了对强干扰噪声图像中的字符串进行实时的检测定位,该文提出了一种基于边缘的字符串定位算法,它引入了边缘密度图和边缘连接强度两个新的概念。该算法首先通过对边缘密度图进行投影分析进行自顶向下的粗定位,然后在此基础上利用垂直边缘的连接强度进行自底向上的精确定位。新算法有效地克服了噪声的影响,运算复杂度低,因而能够实现对强干扰噪声图像中的字符串的实时定位。采用该算法对集成电路芯片图像中的编号字符串进行定位,实验结果证明其在处理强干扰噪声图像时是有效的。An edge-based algorithm was developed for real-time text string location in images with significant noise. The method uses an edge density graph and the edge linkage intensity. The algorithm first makes a top-down coarse location using the projection analysis of the edge density graph and then makes a bottom-up precise location based on the vertical edge linkage intensity. The algorithm effectively overcomes the influence of noise and has low computation cost, so it can be used for real-time text string location in images with noise. The algorithm has been adopted to locate serial number strings in wafer images; the experiment results prove its effectiveness.

关 键 词：文字信息处理 字符串定位 边缘密度图 边缘连接强度 

分 类 号：TP391.1[自动化与计算机技术—计算机应用技术]