基于有效鉴别特征抽取的有限汉字集识别

Recognition of Limited Chinese Character Set Based on Optimal Discriminant Features Extraction

在模式识别领域 ,基于 Fisher鉴别准则函数的最佳鉴别矢量集技术有着重大的影响 .根据统计不相关性最优鉴别变换 ,提出了在汽车牌照二值化图像中抽取汉字字符的有效鉴别特征方法 ,并将其应用于沪宁高速公路收费口处实拍的车牌图像库中 ,其识别率达到 97.13%.实验结果表明 ,此文所提出的牌照图像中汉字字符集的有效鉴别特征方法是有效的 .

Optimal discriminant vectors based on Fisher's discriminant criterion have significant influence on the area of pattern recognition. In this paper, we propose a method to extract uncorrelated optimal discriminant features of Chinese characters in binarized automobile license plate image according to the uncorrelated optimal discriminant transform. The method is applied to recognize Chinese characters in automobile license plate database of Shanghai-Nanjing highway. The recognition rate is 97.13%. Experimental results show that the uncorrelated optimal discriminant features of Chinese characters in automobile license plate image is effective.