关键词: 维吾尔文, 二值图像, 性能比较, Sobel算子, 特征融合

Abstract: This paper introduces four classical edge extraction algorithms of handwriting. The analysis of the experimental result shows that Sobel operator in edge extraction of Uyghur handwriting is better. It proposes a text-independent method of handwriting identification based on feature fusion. Feature fusion includes mesh-window microstructure feature and curvature-direction feature. This method adopts probability density distribution to count the mesh-window microstructure features and the curvature-direction features. The identification effect reaches the expected value. It introduces some new image processing technology, and puts forward the improvement and the fusion of the classical algorithms and new processing technology applied to the Uyghur handwriting identification will be the next step major job.

Key words: Uyghur, binary image, performance comparison, Sobel operator, feature fusion