自适应字符切分及提取算法研究  

Study of adaptive character segmentation and extraction algorithm

作　　者：金海燕[1] 夏婷[1] 王彬[1] 

机构地区：[1]西安理工大学计算机科学与工程学院,陕西西安710048

出　　处：《西安理工大学学报》2016年第4期399-402,415,共5页Journal of Xi'an University of Technology

基　　金：国家自然科学基金资助项目(61472204;61272283)

摘　　要：在字符识别技术日趋成熟的现状下,单个字符的正确切分及提取已经成为制约字符识别精确度的关键因素。本文针对二手车发票上印刷体的日期数字(阿拉伯数字),对图像二值化处理后,采用垂直方向投影和轮廓特征两种策略进行自适应字符切分及提取。实验结果表明,该方法提高了从图像中定位出来的字符串的切分率,并保证了字符切分和提取的正确率平均达到99%。In the current the situation of character recognition technology is becoming more and more mature,and the correct segmentation and extraction of single character has become a key factor to control the accuracy of character recognition.This paper focuses on second-hand car invoice printing of digital date（Arabic numerals）,with both vertical projection and contour feature combination strategies for adaptive character segmentation and extraction being conducted after image binarization processing.Experimental results show that using the proposed method can improve the segmentation rate of the string from the image,ensuring that the average accuracy rate of the extracted image can reach 99%.

关 键 词：字符切分及提取 字符识别 图像二值化 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]