A hyperspectral image rapid classification method based on multi-GPU cooperative interaction data stream organization

Abstract

The invention discloses a hyperspectral image rapid classification method based on multi-GPU cooperative interaction data stream organization, and the method comprises the following steps: reading hyperspectral image training data and test data, and binding the data into a page locking memory; Calculating a maximum likelihood probability matrix through a sparse polynomial logistic regression rapidcalculation method based on multi-GPU stream synchronization, and extracting spectral information of the hyperspectral image; Using a weighted Markov field space prior rapid calculation method basedon multi-GPU variable division to extract hyperspectral image space information, and calculating a marginal probability matrix; And obtaining a prediction label of the test data according to the obtained marginal probability matrix, and obtaining a final classification result. According to the method, the hyperspectral image spectral information is utilized, the spatial information is also utilized, and the classification precision is improved.

Description

technical field [0001] The invention belongs to the field of remote sensing image processing, and in particular relates to a hyperspectral image rapid classification method based on multi-GPU cooperative interactive data flow organization. Background technique [0002] Hyperspectral remote sensing images have the characteristics of high spectral resolution, large number of bands, and map-spectrum integration. It contains not only the spatial information of the entire ground object, but also the spectral information of each pixel. It is a powerful force for people to understand and transform the world. Technical Support. Hyperspectral image classification is an analysis method to describe the types of ground objects. Specifically, in order to distinguish various ground objects contained in the image, according to the various information contained in the image, each pixel is determined according to its characteristics. The corresponding feature name. Initially, researchers o...

AI Technical Summary

Problems solved by technology

[0004] The purpose of the present invention is to provide a hyperspectral image rapid classification method based on multi-GPU collaborative interactive data flow organization, to solve the problems of low classification accuracy and low classification efficiency of existing hyperspectral image classification algorithms

Images