【作者】 芦颖； 倪维平； 严卫东； 边辉；

【Author】 Lu Ying;Ni Weiping;Yan Weidong;Bian Hui;Northwest Institute of Nuclear Technology;

【机构】 西北核技术研究所；

【摘要】 针对目标分解中极化方位角的随机分布和体散射的过估计所导致的全极化合成孔径雷达(POlSAR)图像分类结果的错分问题,构造了一种基于改进极化目标分解的POlSAR图像分类处理方法。首先,对POlSAR图像的相干矩阵进行去极化方位角处理,消除极化方位角对目标的影响;其次,对目标进行修正体散射分量的Freeman分解得到能较好的区分不同的散射地物的三种散射功率分类;最后,采用模糊C-均值(FCM)方法对归一化的散射功率进行分类处理。实验结果显示,在对极化分解之前去极化方位角可以有效降低分解结果中的负功率问题,在Freeman分解中修正体散射分量可以明显的抑制体散射功率的过估计问题,同时有利于增强POlSAR图像分类精度。更多还原

【Abstract】 The randomly distributed orientation angle and over-estimating of volume scattering affects the Polarimetric Synthetic Aperture Radar(POISAR) image target classification quality.In this paper we present a novel POISAR image classification method based on modification target decomposition.First,coherence matrix of POISAR image is deorientationed;then apply Freeman decomposition which modify volume scattering to achieve the initial three components classification;Lastly,use the Fuzzy C-mean(FCM) clustering method to obtain the final image classification.The experiment results show that the deorientation procedure before target decomposition can restrain over- estimate of volume scattering and enhance double-bounce scattering in the decomposition output.This method can avoid negative power in the decomposition output and advance the performance accuracy of classification.更多还原