Binary MRF Model and Multi-Scale Texture Segmentation of Multi-band Remote Sensing Image

A binary Markov random field model is presented for multi-scale texture segmentation of multi-band remote sensing image.An one-time multi-class segmentation problem is converted into a continual binary segmentation problem.At the same time,we scale of 2-d image space and scale of spatial space are synthetically considered,and we got multi-scale texture segmentation is gotten with the improvement of efficiency.