Abstract:

For reducing the impact caused by speckle noise on classification results,a new classification algorithm is proposed for polarimetric synthetic aperture radar (POLSAR) images based on mean-shift and multi-scale Wishart Markov random field. The mean-shift algorithm is used to get the initial classification for the coarsest scale, and then a Markov random field is introduced to achieve the classification result. The classification result on a coarser scale is employed as the initial classification of the nearest finer scale. Meanwhile, the Wishart distribution is employed to model the observed field, and then the iterative conditional modes (ICM) algorithm is adopted to implement the maximum a posteriori estimation of pixel labels for each scale. The classification result of the finest scale is the final classification result for POLSAR images. The algorithm is described in detail and the contrastive experiment is done using AirSAR L band and E-SAR polarimetric images. The experiment result indicates that the proposed method could get higher accuracy of classification than the classical algorithms.