摘要
介绍了一种基于多重隐马尔克夫模型(Multiple HMM——MHMM)的手写体汉字识别新方法。该方法首先提取基于区域投影变换形成的边界链码特征,对每个汉字建立4个HMM,通过等比重综合方法将4个分类器的计算结果进行综合,从而得到识别结果。实验结果证明该方法较传统的HMM具有更高的抗干扰能力和更高的识别率。
A new method of handwritten Chinese character recognition using multiple hidden Markov model is presented. Combined with the contour chain codes by sub - region projection transform, this method firstly constructs 4 HMMs for each Chinese character, and synthesizes the results of the 4 HMMs by equal proportion method, and then gets the final recognition result. Experimental results demonstrate that this method has more anti -jamming ability and higher recognition rate than the traditional hidden Markov model.
出处
《常熟理工学院学报》
2007年第2期85-90,共6页
Journal of Changshu Institute of Technology
关键词
隐马尔克夫模型
手写汉字识别
投影变换
hidden Markov Model
handwritten Chinese character recognition
projection transform
