摘要
从文本图像中提取局部特征采进行文字种类识别,对图像的质量要求较高,而提取文本图像的整体特征进行识别,则能够克服了这个问题。分析中、英文文本图像的纹理特征,通过Gabor滤波器分析提取全局特征,再用支持向量机的方法进行文种分类识别。实验结果表明,该方法可以很好地区分含噪文本图像中的中文和英文。
Using local texture analysis based approach to make script recognition need fine quality text image, but global texture analysis based approach doesn＇t need so. A recognizing method for script recognition is introduced on the basis of the analysis of Chinese and English texture features. It is a global texture analysis based approach using gabor filters, and use SVM to make script recognition. The experimental results are very promising.
出处
《计算机工程与设计》
CSCD
北大核心
2007年第21期5150-5152,共3页
Computer Engineering and Design
