在分析了极化干涉技术的发展现状和发展趋势的基础上,从极化干涉的特殊性入手,分析了干涉相干系数在极化干涉分类中的作用,并且系统分析了基于相干最优优化后的相干系数进行极化干涉分类过程的原理,总结了现有极化干涉分类算法在阈值选择方面存在的不足,阐述了硬阈值方法和软阈值方法的基本原理和它们之间的区别,最后提出基于软阈值方法的改进A1/A2非监督极化干涉分类算法,解决了极化干涉分类现有的分类空间中的阈值问题。实验表明这种方法能够得到更好的分类结果,做出了改进。
Based on analysis on development status and trend of polarization interference technology,this paper starts with the analysis of the particularity of PolInSAR;And then analyzes the role of coherence in PolInSAR classification process;Analyzes the principle of PolInSAR classification method based on the coherence optimization process;Summarizes the shortcomings in the choice of threshold in the classification process,and represents the principle and difference between the hard threshold and soft threshold;Finally proposes an improved A1/A2 unsupervised PolInSAR classification method based on the soft-threshold method,which has solved the problem of threshold choosing in classification.Experimental results have shown that this method can get better classification results than the existing A1/A2 classification algorithm.