摘要
提出了一种新的应用于大规模复杂图像分割的谱聚类方法,该方法通过均匀采样获取图像的较小模式,通过快速卡通—纹理分解模型分解图像,分别获取图像的光谱和纹理特征,然后通过Nystrm谱聚类算法确定采样图像的划分,最后利用其结果,依据一种综合了K近邻以及随机选择思想的估计规则确定原图像的最终划分。大规模合成纹理图像及自然图像的分割实验验证了该方法的可行性及有效性。
This paper proposed a new spectral clustering method for large-scale complex image segmentation.This approach firstly obtained the smaller model of an image by uniform sampling,got the spectral and texture features of the smaller image above on through fast cartoon-texture decomposition models.Then determined the segmentation of the smaller image by Nystrm spectral clustering.Finally,used the above segmentation to estimate the final classification of the original image based on the rule containing the idea of K nearest neighbor and random selection.Some large-scale synthesized textured images and nature images were used for testing.The experimental results show the feasibility and effectiveness of the proposed method.
出处
《计算机应用研究》
CSCD
北大核心
2011年第5期1994-1997,共4页
Application Research of Computers
基金
国家自然科学基金资助项目(40671133)
中央高校研究基金资助项目(GK200902015)
关键词
谱聚类
大规模图像
卡通—纹理分解
采样
估计
spectral clustering
large-scale image
cartoon-texture decomposition
sampling
estimate
