To effectively fuse the high-dimensional features and make full use of the contribution of different classifiers to image classification, we propose a polarimetric synthetic aperture radar (PolSAR) image classification method fusing SVM iterative classification and JSRC decisions, SVM-JSRC for short. This method combines the idea of semi-supervised classification and uses SVM classifiers for multiple iterative classification, which is used to expand the training sample set and reduce the influence of insufficient PolSAR image label samples. Since various characteristics would contribute differently to image classification, the extended training sample set was used to extract the high-dimensional polarization and texture features. And then constructing the feature dictionaries, respectively, which are used to make joint sparse representation classification (JSRC) decisions. To get classification results, SVM classification and JSRC decisions are combined by using Dempster-Shafer (D-S) evidentiary theory. The efficacy of the suggested strategy is demonstrated by experiments using actual PolSAR images.