基于非线性概率松驰过程的汉字识别后处理方法

THE POST-PROCESSING METHOD OF CCR BASED ON NON-LINER PROPABALITY RELAXATION PROCESS

本文通过对松驰方法的分析与改进,提出了一种基于非线性概率松驰过程的汉字识别后处理方法。同常用的词匹配方法相比,该方法使用了更多的汉字及汉语词汇信息和初级识别结果所提供的信息并把这些信息有机地结合起来。另外,该方法还是一种高度并行的方法,运行速度可以很快。

Based upon the analysis and improvement of the relaxation method,this paper proposeda post-processing method of CCR based on non-liner propabality relaxation process.Comparing with theword matching method,this method not only uses more information of Chinese Characters and words,butalso uses the information given by the initial recognition results.Moreover,this is a highly parallel methodand the working speed can be very fast.