|
|
|
| SAR Image Segmentation Combining Possibilistic C-Means Clustering and Spatial Information Optimized with Immune Clonal Algorithm |
| Tian Xiao-lin; Jiao Li-cheng; Gou Shui-ping |
| Institute of Intelligent Information Processing, Xidian University, Xi’an 710071, China |
|
|
|
|
Abstract Possibilistic C-Means (PCM) clustering algorithm exhibits the robustness to noise, but the spatial information is not considered in this algorithm. Due to the effect of speckle in Synthetic Aperture Radar (SAR) images, the serious inaccuracies with segmentation can be resulted by using the PCM algorithm. A robust segmentation algorithm based on an extension to the traditional PCM algorithm is proposed in this paper. The relative location information and intensity information of neighboring pixels are introduced into the objection function of the PCM algorithm. The values of these information are determined by previous clustering result. The degree of influence of these information on clustering is optimized with Immune Clonal Algorithm (ICA), so the degree of influence is adjusted adaptively. Meanwhile, the clustering results of the PCM algorithm are optimized. In the paper, synthetic image and real SAR images are segmented to demonstrate the superiority of the proposed algorithm. The experimental results show that the proposed algorithm is insensitive to the initial segmentation result and improves the segmentation performance dramatically.
|
|
Received: 11 June 2007
|
|
|
|
|
|
|
|
|
2008, Vol. 30 
