Research on Hyperspectral Remote Sensing Image Classification Method based on CS and SVM Improved Algorithm

Sheng Cang1,2, A'chuan Wang1

1Northeast Forestry University, Harbin, 150000, China

2Heilongjiang International University, Harbin, 150000, China

Abstract: Hyperspectral remote sensing image classification is a main application of hyperspectral. Due to the large number of bands and large amount of data in hyperspectral remote sensing image, the classification accuracy is not high and the classification time is long in the process of hyperspectral classification. In this paper, the methods of supervised classification and unsupervised classification of hyperspectral classification are studied comprehensively, and an improved SVM classification method is proposed to solve the problems existing in the classification process of SVM. Although this method can improve the accuracy of classifica-tion, it cannot solve the problem of large data processing. Therefore, this paper applies the compressed sens-ing theory to the improved SVM method, which can realize the classification of hyperspectral remote sensing images on the basis of sparse basis, thus not only increasing the classification accuracy, but also reducing the computational ｌｏａｄ. Simulation results show that the algorithm has better classification characteristics.

Keywords: Compressed sensing; SCV. hyperspectral; Sparse matrix