脱机手写签名图像的细化研究

Study on thinning of off-line handwritten signature image

针对脱机手写签名图像的细化存在的问题如细化后转折处出现的分叉笔画、两笔画不垂直相交部分的畸变,提出了新的细化算法。此算法采用串行处理的方式,首先对签名图像进行边缘提取,再针对边缘像素点进行可删除条件的判断,经过几次循环,最后形成细化图像。实验结果表明,此方法能较好地解决以上提到的问题,尤其是分叉笔画。

A new thinning algorithm is proposed to solve such problems existing in thinning off-line handwritten signature images as branch stroke's appearing in the turn and abnormal changes in the not vertical crossing part of two strokes. A serial method is used in this algorithm. Firstly, the contour of signature image is extracted. Then it's determined whether the contour pixels can be removed. Finally a thinned image is generated. Experimental results show that the problems previously discussed especially branch stroke can be well solved by this algorithm.