消噪和数学形态学结合的字符图像预处理算法  

Preprocessing Algorithm of Character Image by Assembling Denoising Algorithm and Mathematical Morphology

作　　者：陈利娟[1] 徐利华[1] 

机构地区：[1]西南交通大学物理科学与技术学院,四川成都610031

出　　处：《现代电子技术》2009年第22期110-111,118,共3页Modern Electronics Technique

基　　金：西南交通大学基础研究基金资助项目(2007B18)

摘　　要：噪声污染字符图像的预处理是字符识别技术中一项关键技术,字符识别的准确度直接取决于字符图的质量。基于消噪和数学形态学的基本原理,提出了一种预处理算法。仿真实验验证了所提出新算法的有效性和可行性。这种算法不仅能消除图像中的噪声,而且还有效地修复了笔划断裂的字符。新算法解决了识别笔划缺损字符的识别难点,从而间接地提高了字符识别技术的准确度。Preprocessing of noised- character image is a key technique in character recognition. The accuracy of character recognition depends on the quality of character image. According to the fundamental principle of denoising algorithm and mathematical morphology,a preprocessing algorithm is made. Simulation experiments verify validity and feasibility of the new algorithm. Not only noise can be eliminated, but also the stroke parted characters become connectivity. The recognition difficulty of the stroke parted characters may be resolved by using this algorithm. Consequently,the accuracy of character recognition technique is improved indirectly.

关 键 词：数学形态学 加权平均 均值滤波 中值滤波 形态学滤波 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]