支票图像中打印体字符的分割和识别

Segmentation and Recognition of Printed Character in Check Image

和通常文档中的打印体字符分割和识别问题相比，支票图像中的打印体字符的分割和识别有其一定的特殊性。针对这种特殊性，该文提出了一套简单实用的算法。对于分割问题，主要是采用投影的方法解决印章、边框和底纹的干扰问题，对于字符识别，主要是采用了基于模板和融合的方法，从而有效地解决了上述问题，取得了比较满意的分割效果和较高的识别率。

Compared with the problem of segmentation and recognition in common document, the problem of segmentation and recognition in check image has certain particularity. In the view of this particularity, this paper presents a set of simple and practical method. For the segmentation, it mainly adopts the projection method to overcome the interference of seal, frame and shading. For the recognition, the method based on pattern and fusion is used, which solves the above problems effectively. The satisfied segmentation result and high recognition rate is obtained.