摘要
在传统的图像块分类方法中,利用局部方差可以将图像块划分为平滑块和非平滑块,但是对于非平滑块中包含的边缘块和纹理块,则不能有效进行区分。针对这一问题,提出一种基于结构稀疏度的图像块分类方法,根据图像块与其邻域的其余图像块之间的相似程度对图像块的局部特征进行辨识。仿真实验结果表明,该方法可以对平滑块、边缘块,以及纹理块进行有效区分。
In the traditional image patch classification method, the image patches can be divided into smooth patch and non-smooth patch using the local variance. However, for the non-smooth patch, the edge patch and texture patch cannot be distinguished effectively. In view of this problem, proposes an image patch classification method based on structure sparsity. The local features of image patches are identified according to the similarity between the current image patch and the rest of the image patches in its neighborhood. Simulation experiment results show that the method can distinguish the smooth patch, the edge patch, and the texture patch effectively.
基金
2014年运城学院教学改革项目(No.JG201429)
2016年运城学院教学改革项目(No.JG201628)
关键词
结构稀疏度
相似度
图像块
分类
Structure Sparsity
Similarity
Image Patch
Classification
