极化SAR数据的细节保持分割

提出一种保持场景局部细节的极化合成孔径雷达数据非监督分割方法.该方法将传统的基于极化目标分解的极化SAR分类方法与基于马尔可夫随机场的分类方法相结合,利用迭代条件模型方法得到分割结果.为保持细节特征,同时又能对各向同性区域进行平滑,利用总功率(span)图像的场景均匀性对分类结果进行修正.实验结果表明,与已有方法相比,该方法在细节保持方面有一定改进.

Detail-preserving segmentation of polarimetric synthetic aperture radar data

An unsupervised detail-preserving segmentation method of polarimetric synthetic aperture radar(POLSAR) data is proposed.This method combines the classical POLSAR data classification method based on polarimetric target decomposition and the classification method based on Markov random field(MRF).An Iterated conditional modes(ICM) method is used to obtain the segmentation results.In order to preserve details and smooth the homogeneous regions,the classification result is modified by using the scene homogeneity of the total power(span) image.Experiment results indicate that the proposed method achieves some improvements for detail preservation compared with existed methods.