基于连通域的扭曲中文文本图像快速校正方法  

Fast correcting method based on connected components for warped Chinese document images

作　　者：曾凡锋[1] 郭正东[1] 王战东 

机构地区：[1]北方工业大学信息工程学院,北京100144

出　　处：《计算机工程与设计》2015年第5期1251-1255,共5页Computer Engineering and Design

基　　金：国家自然科学基金项目(61371142);国家科技支撑计划基金项目(2012BAH04F03);北京市自然科学基金项目(4132026);北京市科技创新平台基金项目(PXM2013_014212_000011)

摘　　要：针对扭曲中文文本图像文字识别率不理想这一问题,提出一种基于连通域的文本图像快速扭曲校正方法。根据汉字结构特征合并连通域,实现切分文字;利用就近聚合文字的方法定位文本行,按行垂直校正每个文字位置,获得被校正的图像。实验结果表明,该方法校正速度快,对严重扭曲的中文文本图像能取得较好的校正效果,校正后图像的OCR识别率明显提高。Character recognition rate of OCR （optical character recognition）processing is not satisfactory for warped Chinese document image.To resolve this problem,a fast distortion correcting method based on connected components was proposed. First,the connected components were combined together according to the Chinese character structure characteristics.Next,the Chinese characters were segmented one by one according to the combined connected components.After that,the text lines were identified based on the nearest aggregation method.Then,the vertical positions of the segmented characters were corrected ac-cording to every text line.As a result,a well corrected document image was obtained.Experimental results demonstrate that this correcting method is fast and can segment the Chinese character accurately.The OCR rate of the corrected images can be sig-nificantly improved.Even for the obviously distorted Chinese document images,this method can achieve better results.

关 键 词：中文文本图像 扭曲图像 连通域 文字切分 就近聚合 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]