摘要
相似字多是造成汉字识别误识率和拒识率高的主要原因之一,该文提出了一种基于动态特征选择的相似字识别方法,其识别过程从初始提取全局特征开始,然后逐步动态地、递归地加入更精细的局部特征以提高识别的判决力,直至识别结果满足判决条件为止。这种方法不需要人工确定相似字组,而且能自动选择相似字间区别最大的部分空间,构成新的特征向量。通过实验验证,该方法使相似字的识别率有了显著提高,证明了该方法的有效性。
A large amount of similar Chinese characters is one of the main reasons to cause the high reject rate and substitution rate, This paper proposes an innovative similar character recognition method based on dynamical feature selection. The process includes originally abstracting global feature vectors, progressively, dynamically and recursively adding more accurate local feature vectors to improve the ability of recognition and finally achieving the result that satisfies the conditions. In this way, it is not necessary to decide similar characters group manually since it can automatically choose the largest space in the difference of similar characters to construct new feature vectors. The efficiency of this method is proved by the experiments that effectively improved the recognition rate of similar characters.
出处
《计算机工程》
EI
CAS
CSCD
北大核心
2006年第17期10-11,18,共3页
Computer Engineering
基金
湖北省科技攻关基金资助项目(2003BDST004)
关键词
相似字
动态特征选择
部分空间
Similar Chinese characters
Dynamical feature selection
Partial space
