基于核函数的联合稀疏表示高光谱图像分类

为了充分利用高光谱图像邻域像元间的相似性与独特性这一特征信息,提出了一种基于核函数的联合稀疏表示分类方法(kernel joint sparse representation classification, K-JSRC)来提高高光谱图像的分类精度。该方法通过一种改进的核函数对每个待测中心像元的所有邻域像元自适应的予以不同权重来测量待测中心像元与邻域像元的相似度从而得到不规则的最优邻域窗口。在Indian Pines和University of Pavia两个高光谱数据集上的实验结果表明,提出的分类算法对高光谱图像进行了很好的分类并且其分类精度优于同类算法。

Joint sparse representation of hyperspectral image classificationbased on kernel function

In order to make full use of the similarity and uniqueness of hyperspectral image neighborhood pixel, a joint sparse representation classification (K-JSRC) method based on kernel function is proposed to improve the classification accuracy of the hyperspectral image. The method uses an improved kernel function to measure the similarity of the neighborhood of the measured cell to the neighborhood pixel by adapting all the neighborhood pixels of each neighborhood of the measured cell to obtain the irregular and excellent neighborhood window. The experimental results on the two hyperspectral data sets of Indian Pines and University of Pavia show that the proposed classification algorithm has a good classification of hyperspectral images and its classification accuracy is superior to the similar algorithms.