A semi-supervised col laborative classification for hyperspectral remote sensing image with com-bination of cluster feature and SVM is proposed.The frame of the proposed method combines kernel-spectral fuzzy c-means and semi-supervised SVM to improve the classification accuracy,through making ful l use of the advantages of classification and clustering.In detai ls,ClusterLoss,ClassConsistent,classi-fication difference and sample difference are created to bui ld the col laborative classification frame,which can make the best of l imited labeled samples and lot unlabeled data.This approach can minimize the cost of acquisition of labeled samples and in some degree solve the problem that support vector increases l ine-arly with the number of training samples.Experimental results show that classification accuracy of the pro-posed method is more effective than that of semi-supervised SVM.%提出一种聚类特征和 SVM 组合的高光谱影像半监督协同分类方法。利用构建的协同分类框架能够将 KSFCM 聚类算法与半监督 SVM 分类器相结合，同时利用聚类和分类优势，提高分类器的分类准确率。其中，通过聚类损耗函数、分类一致函数、分类差异性、样本差异性4个指数用以构建协同分类框架，以充分利用少量类标签样本信息，避免高光谱类标签样本获取困难问题，在一定程度上解决 SVM 支持向量随着训练样本增加而线性增加的问题，从而寻求最佳分类结果。试验结果表明，本文所提方法得到的分类精度优于直接利用 SVM 进行半监督分类。
展开▼
