计算机自动识别字符图像的二维滤波算法  

Two-Dimensional Filter Algorithm for Automatic Recognition of Character Images by Computers

作　　者：杜媛[1] 

机构地区：[1]西安职业技术学院动漫软件与教育学院,陕西西安710077

出　　处：《西安职业技术学院学报》2017年第1期37-40,共4页Research on Vocational Education in Xi'an Vocational and Technical College

摘　　要：提出了基于二维滤波Gabor与机器视觉的字符识别算法，从图像预处理和特征提取分析这两个模块展开，就是为了解决字符识别算法在噪声、旋转下出现识别不准确的问题．首先，搭建视觉采集硬件平台，采集原始字符图像．然后通过邻域像素灰度信息迭代，有效降低噪声，以提高字符图像质量，通过霍夫直线检测完成图像旋转角度计算和图像校正，达到准确定位字符区域的目的；其次，通过二维滤波Gabor得到字符图像的纹理特征，采用欧式距离构建分类标准，达到准确识别字符的目的．实验测试数据表明：与当前识别机制相比，在面对噪声与旋转干扰条件下，二维滤波算法具有更高的识别准确率与鲁棒性．Character recognition algorithm on computer vision is proposed in this paper from the image pre-processing and feature extraction and analysis of the two modules, so as to solve the character recognition algorithm in noise, identify inaccurate problem under text rotation. Firstly, creating visual acquisition hardware platform and collecting the original character image. Then, by 2d Gabor filter to get the texture feature of character image, the classification criteria of Euclidean distance is used to construct and achieve the purpose of accurate identification character. Experimental test data show that compared with the traditional recognition mechanism, this mechanism in terms of recognition has higher accuracy and robustness.

关 键 词：字符识别 计算机视觉 二维滤波Gabor 霍夫直线检测 

分 类 号：TP391.1[自动化与计算机技术—计算机应用技术]