LSSVM算法在极化SAR影像分类中的应用

Research on Polarimetric SAR Image Classification Based on Least Squares Support Vector Machine

最小二乘支持向量机(LSSVM)是针对标准支持向量机(SVM)算法训练时间长的问题而提出的一种改进算法。针对SVM算法在极化SAR影像分类时存在效率较低的问题,以目标分解理论为基础,对LSSVM算法应用于极化SAR影像分类的有效性进行了研究。结果表明,对于极化SAR影像分类,LSSVM算法与SVM算法的分类精度相当,但时间效率远优于SVM算法,并且对参数的调整也具有更好的稳定性,同时泛化能力良好。

The LSVM algorithm was put forward for the inherent shortcomings of long training time that standard support vector machine algorithm has.In order to solve the efficiency problem that the SVM algorithm has in supervised classification for the polarimetric SAR images,based on the target decomposition theory,we used the LSVM algorithm to do supervised classification for the polarimetric SAR images and tested the effectiveness of this algorithm for the classification of the polarimetric SAR images.The experiment shows that,in the polarimetric SAR image classification applications,the LSSVM algorithm can be quite the classification accuracy with SVM,and though the efficiency and accuracy comparison find that the LSSVM algorithm has a faster speed and better stability,and has a better generalization ability.