基于统计特征的维吾尔文笔迹鉴别研究

针对目前的笔迹鉴别研究只是单一针对纹理特征和概率分布密度．提出一种采用边缘方向分布特征和LBP纹理特征相融合的笔迹鉴别方法。该方法利用概率密度分布思想从边缘轮廓图像中提取边缘方向分布特征，使用直方图向量提取LBP纹理特征。并且采用几种距离公式进行相似性度量。所提方法简单可行并很好地描述了维吾尔文的书写习惯和特征，取得较好的鉴别结果。

Research on Uyghur Handwriting Identification Based on Statistical Feature

For the current handwriting identification study is only for texture feature or probability density distribution, proposes a handwriting identification method based on feature fusion which involved edge orientation distribution and LBP texture feature. The proposed method extracts edge contour image from the original image. Furthermore, extracts edge direction distribution featrues using a method of probability density distribution and extracts LBP textrue features using histogram vector. Uses a variety of distance formulas to measured similarity. With a relative strong practicability, the presented method is feasible and applicable and well described handwriting feature and style of Uygur letter.