基于多方向特征融合的维吾尔文笔迹鉴别技术

在笔迹图像中格线和噪音的去除、细化等预处理基础上，结合维吾尔文笔迹结构和书写风格，提出了一种基于四维笔划方向特征的笔迹鉴别技术。为了进一步提高其鉴别率，还将方向特征与较成熟的基于倾斜度的另一种方向特征进行了融合，取得了较好的实验结果。具体实施过程中，还对比分析了不同的特征距离度量方法对鉴别率的影响，确定加权欧式距离为最佳度量方法。

Uyghur handwriting identification technology based on combining of multi-directional features

On the basis of suppression of grid lines and noises in handwritten paper images and thinning, according to the Uyghur handwriting structure features and writing styles, this paper presents a naive handwriting identification technique based on the four-dimensional directional features of the subset strokes. In order to further improve the identification rate, it fuses the directional feature presented in this paper with another often used slope computing based directional feature, then achieves a better handwriting identification rate. In the implementation process, this paper also comparatively analyzes the effects of different distance measurement methods to the identification rate, and selects the weighted Euclidean distance as the best measurement method for the case of this paper.