基于识别结果反馈信息的闭环联机字符识别系统

文章提出了一种基于识别结果反馈信息的闭环联机字符识别系统。通常的识别系统只是执行了从输入数据空间到识别结果空间的单向映射,可以看作为一种开环识别系统;而文中提出了一种闭环识别系统:它在完成初次的识别之后,还将识别结果信息反馈给系统中的规则库;规则库依据此反馈信息触发相应规则驱动识别器对输入字符进行再次识别,从而得到最终的识别结果。实验结果表明,在抽取的特征和识别方法相同的条件下,闭环识别系统的识别率将高于开环识别系统的识别率。

A Closed Loop On-line Character Recognition System Based on Feedback Information of Recognition Results

In this paper,a closed loop on-line character recognition system based on feedback information of recogni-tion results is proposed.The conventional recognition system may be regarded as an open loop recognition system,which only executes a mono-direction mapping from input data space to recognition result space.However,the proposed system in this paper is a closed loop recognition system:after obtaining the initial recognition results,the system further feed back recognition results to the rule database of the system.According to the feedback information,the rule database touches up a corresponding rule to drive the recognizer to re-recognize the input character to get the final recognition results.The conducted experiment results have shown that under the same extracted features and the same recognition method conditions,the recognition rate of a closed loop recognition system is higher than that of an open recognition system.