Spectral-spatial Hyperspectral Image Classification Based on Sparse Representation and Edge Preserving Filtering

Hyperspectral remote sensing technology is rapidly developing and has been widely used in many industries; classification is one of the most important research contents of hyperspectral data processing. The combination of spatial and spectral information is applied in the classification of hyperspectral images which is known to be an effective way to improve classification accuracy. In this paper, the proposed framework consists of the following three steps. Firstly, this research, in the premise that the value of allowable error is optimal, adopts the pixel classifier based on sparse representation to classify the hyperspectral image. Then, the resulting classification map is represented as multiple probability maps, and edge-preserving filtering is conducted on each probability map. Finally, according to the filtered probability maps, the class of each pixel is selected based on the maximum probability. The experimental results show that the method has a good classification result, and can be widely used in all walks of life.