A novel unsupervised segmentation for remote sensing image using MRF
Jiajing Wang1, 2, Shuhong Jiao1, Zhenyu Sun3
1Faculty of Information and Communication Engineering, Harbin Engineering University, No.145, Nan Tong Road Harbin, Heilongjiang, China
2No.92677 Unit of PLA, Dalian, Liaoning, China
3No.91550 Unit of PLA, Dalian, Liaoning, China
The image segmentation is the basis of image interpretation in remote sensing applications and plays vital role in image analysis. The Markov Random Field (MRF) approach is widely studied for use in segmentation of remote sensing image, which is an important extraction technique in recognition problems. This paper presents an unsupervised segmentation method for remote sensing image using the MRF. A novel neighbourhood system for the energy function has been proposed, the segmentation of remote sensing image and the optimization process of the parameters are performed simultaneously for the unsupervised segmentation in iterative condition. The experimental results on Synthetic Aperture Radar (SAR) images show that the proposed method performs better than the conventional Bayesian segmentation methods.