摘要： 针对基于纹理分析的全局特征提取受文本组合或字符连接影响较大，导致特征提取不稳定的现象，以及现有局部特征提取方法存在的不足，提出一种基于局部结构分割构建Codebook的维吾尔文文本无关笔迹鉴别方法。该方法根据笔迹图像的像素值提取维吾尔文笔迹样本中具有代表性的轮廓，生成描述书写者书写风格的Codebook并进行相似性度量，从而达到笔迹鉴别的目的。实验结果表明，该方法对于维吾尔文笔迹是一种简单、可行，具有较高识别率的笔迹鉴别方法。

关键词: 笔迹鉴别, 维吾尔文笔迹, 轮廓提取, 轮廓细化, 轮廓分割, Codebook构建

Abstract: Due to the great impact on stable feature extraction in global texture analysis for feature extraction by text or character combinations, and existing local feature extraction methods’ shortages, this paper presents a text-independent writer identification method for Uyghur handwritings based on local structure segmentation and Codebook construction. The proposed method extracts representative contours from handwriting samples according to pixel values, generates a Codebook of specified writer to depict writer’s writing style and uses Codebook to measure the similarity of different writers to achieve writer identification. Experimental results demonstrate that the presented method is a feasible and applicable writer identification method with favorable performance.

Key words: handwritings identification, Uyghur handwritings, contour extraction, contour thinning, contour segmentation, Codebook construction

中图分类号: 

• TP391.41