Abstract: This paper proposes a text-independent method of writer identification based on sliding-window local contour structure feature.The proposed method depicts all kinds of local contour structure probability distribution in handwritings and uses Chi-square distance to measure the similarity.The experimental results show that this method improves the identification accuracy rate in the HIT-WM open database involving 240 writers.

Key words: local contour, structure coding, probability distribution function, Chi-square distance, writer identification

摘要： 提出一种基于滑动窗口的局部轮廓结构特征的文本无关的笔迹鉴别方法。该方法利用概率分布函数对笔迹中出现的各类局部轮廓形状结构的分布进行了描述，并采用卡方距离度量方法对笔迹进行最终的相似性度量。实验结果表明，在包含240人的HIT-MW中文笔迹库上有效地提高了鉴别正确率。

关键词: 局部轮廓, 结构编码, 概率分布函数, 卡方距离, 笔迹鉴别