Index-guided natural image segmentation
Dongxiang Chi, Ming Li, Ying Zhao, Jing Hu · 2011
Natural image segmentation has been a major research topic in recent years. From the viewpoint of clustering, image segmentation could be solved by Self-Organizing Map (SOM) based methods. In this paper we combine a saliency map with SOM and k-means method (SOM-KS) to segment a natural image. Features of saliency map, intensity and L*u*v* color space are trained with SOM and followed by a k-means method to cluster the prototype vectors. The guidance of an entropy or quantitative evaluation index helps to make a more precise segmentation. Comparison shows that the proposed unsupervised method can achieve better segmentation results, less computational load and no human intervention with the guidance of the entropy index.