【作者】 王强； 齐敏；

【Author】 WANG Qiang;QI Min;School of Electronics Information,Northwestern Polytechnical University;

【机构】 西北工业大学电子信息学院；

【摘要】 针对模糊低品质光学字符识别准确率低的问题,在研究已有光学字符识别算法的基础上,改进优化相关算法,调整识别结构,设计完成一套模糊低品质字符串识别系统。在图像增强方面,提出一种以Canny算子为卷积核,结合Laplace锐化操作,对模糊字符进行图像增强的算法;在文本检测方面,实现了基于MSER算法的文本检测;在模糊字符识别方面,设计了一种两层结构分类器,缩短了识别时间,提高了识别准确率。更多还原

【Abstract】 Aiming at the problem of low accuracy of fuzzy low-quality optical character recognition,based on the research of existing optical character recognition algorithms,this paper improves optimization related algorithms,adjusts the recognition structure,and designs a fuzzy low-quality character string recognition system.In terms of image enhancement,an algorithm that uses Canny operator as the convolution kernel and combines Laplace sharpening operation to enhance the image of fuzzy characters is proposed.In terms of text detection,text detection based on the MSER algorithm is implemented.In terms of fuzzy character recognition,a two-layer structure classifier is designed to shorten the recognition time and improve the recognition accuracy.更多还原