Hyperspectral image compression method based on deep learning and distributed information source coding

Abstract

The invention provides a hyperspectral image compression method based on deep learning and distributed information source coding. The hyperspectral image compression method comprises the following steps: step 1, constructing a hyperspectral image saliency detection deep learning network model; step 2, extracting spectral segment groups and key frames of a to-be-compressed hyperspectral image; 3, extracting the spectral band group local significance characteristics of the to-be-compressed hyperspectral image; 4, obtaining a global saliency mapping graph of the spectral segment group; step 5, obtaining a region of interest of the spectral band group of the hyperspectral image to be compressed; step 6, performing distributed compression on the region of interest of the spectrum segment group;7, obtaining a compressed code of the hyperspectral image. According to the method, the defect that the scene saliency deep representation problem is difficult to solve in the prior art is overcome,and the method has the advantage of accurately compressing useful information; the method overcomes the defect of low hyperspectral image compression efficiency in the prior art, and has the advantageof quickly realizing compression.

Description

technical field [0001] The invention relates to the technical field of image processing, in particular to a hyperspectral image compression method based on deep learning and distributed source coding. Background technique [0002] Hyperspectral images organically combine the spectrum that determines the properties of ground objects with the images that measure the spatial geometric relationship of ground objects, and are widely used in military reconnaissance and national economy and other fields. However, with the continuous improvement of spectral, spatial, temporal, and radiometric resolutions and quantitative depths, the data volume of hyperspectral images is increasing exponentially, and data storage and transmission are facing enormous pressure. How to effectively compress the big data of hyperspectral remote sensing and adapt to the application requirements has become an urgent problem. [0003] Song Juan, Wu Chengke, Zhang Jing and others proposed a method based on ...

AI Technical Summary

Problems solved by technology

This method can save specific areas from damage, but finding specific areas increases the complexity of the algorithm

Images