Feature Extraction of Hyperspectral Remote Sensing in Parallel Computing Research Based on GPU

This paper introduces the basic modes of image parallel processing, and analyzes the parallel processing ability and stream program mode of GPU. At the same time, in view of the hyperspectral remote sensing data due to the large amount of data caused by the operational characteristics of a long time, we study feature extraction of hyperspectral remote sensing algorithms of parallel strategy, especially for the PCA parallel process based on GPU. At last, we conduct a system experiment on hyperspectral remote sensing images to prove the validity of this method.