Hyperspectral image tensor feature extraction based on fusion of multiple spectral-spatial features

Abstract

Aiming at the problem that current hyperspectral image tensorfeature extraction methods cannot make full use of the multiple spectral-spatial features of hyperspectral image, a new hyperspectral image tensor feature extraction method based on fusion of multiple spectral-spatial features is proposed in this paper. Firstly, three-dimensional Gabor wavelets were used to get multiple texture features with different directions and different frequencies, and the multiple shape structural features were obtained by different morphological attribute filters. The tensor features are constructed by combining the spatial feature, multiple texture features and multiple shape structural features. Then, using the local tensor discriminant analysis, the proposed algorithm effectively increases the consistency of the same kind tensors and the difference between different kinds of tensors, which can get lower dimensional tensors consisting of discriminating information and multiple spatial-spectral features. The experiments are performed on the Pavia University and Salinas hyperspectral datasets. Experimental results indicate that the proposed method can maintain the spatial-spectral information and discriminating information, which has better classification accuracy than other algorithms when it is applied to the classification images, and the overall classification accuracies reach 97.74% and 98.03%, respectively. Because of combing the multiple spatial-spectral features, the proposed algorithm can get more effective discriminating features. Thus, the proposed method effectively improves ground object classification accuracy of hyperspectral data and gets the classification map with better spatial continuity.