基于轮廓方向特征的脱机手写中文笔迹鉴别

摘要

提出了一种基于轮廓方向特征的脱机手写中文笔迹鉴别方法.轮廓方向特征是在笔迹轮廓上的每一个像素点周围一个局部的微型网格中获取笔迹轮廓的方向角度和笔迹曲率信息,描述笔迹图像独特的个人手写风格.在获取特征向量之后,使用带权卡方距离计算特征向量的相似度.分别以信封地址图像库中25个书写入的信封地址图像和HIT-MW库中390个书写入的手写中文笔迹图像为实验对象.最佳Top-1鉴别结果达到了95.02％的正确率.%In this paper, contour-directional feature for the issue of text-independent writer identification of offline Chinese handwriting is proposed. The contour-directional feature encodes orientation and curvature information in a local grid around every edge pixel to give an intrinsic characteristic of individual handwriting style. The improved weighted Chi-square metric is applied to measure the similarity of two handwritings. The system is tested on a handwriting database of offline Chinese handwriting involving 390 writers. The best performance of Top-1 identification accuracy is 95.02%.