联合自适应邻域和Bagging的协同表示集成学习方法

Collaborative Representation Ensemble Learning Method Combining Adaptive-shape Neighborhood and Bagging

近年来,协同表示分类(collaborative representation classification,CRC)算法成为高光谱遥感影像分类的研究热点,其中基于bagging的协同表示集成学习算法(bagging-based collaborative representation classification,BagsCRC)利用bagging集成方式有效地提高了基分类器协同表示分类算法的精度。为进一步提升BagsCRC算法的有效性,文章提出了一种联合自适应形状邻域和bagging协同表示集成学习算法(shape-adaptive bagging-based collaborative representation classification,SABagsCRC)。该算法通过构建训练样本和测试样本的自适应形状邻域,进而构建空间信息约束的分类器集成模式。实验采用Indian pines和Washington DC Mall两组高光谱遥感影像,对所提出算法的性能进行了评价。实验结果表明,SABagsCRC算法在分类效果上比BagsCRC算法有明显的提升。

Recently,collaborative representation classification(CRC)has attracted much attention in hyperspectral image analysis.Due to using the bagging ensemble method,bagging-based collaborative representation classification(BagsCRC)achieves better performance.Furthermore,in order to improve the classification accuracy,a novel shape-adaptive bagging-based collaborative representation classification ensemble method(SABagsCRC)for hyperspectral image classification is proposed.The performance is evaluated by using Indian pines and Washington DC Mall hyperspectral remote sensing images.The results show that the classification effect of SABagsCRC is significantly improved than that of BagsCRC.