Efficient construction method of hyperspectral remote sensing data compression and classification model

Abstract

The invention provides an efficient construction method of hyperspectral remote sensing data compression and a classification model which comprises the following specific steps: preprocessing originalhyperspectral data, and segmenting the original hyperspectral data into single hyperspectral pixels; designing the output of the coding part of the full convolution automatic encoder network into binary output to obtain an automatic encoder CAE with efficient spectrum compression, and designing a joint deep learning network CAELR with efficient spectrum compression and rapid classification by combining the CAE with a logistic regression LR classifier; combining CAELR and JPEG2000 to design a method CAELR + JP2 with spectrum and space full-dimension compression and classification; combining the trained CAELR with JPEG2000 to carry out rate distortion optimization so as to realize the optimal rate distortion performance of the CAELR + JP2 under each bit rate; the CAELR + JP2 model designedby the invention effectively improves the compression and classification precision and speed of the hyperspectral data between the satellite-borne hyperspectral sensor and the ground receiving station.

Description

technical field [0001] The invention relates to the fields of aerospace technology and artificial intelligence technology, in particular to an efficient hyperspectral remote sensing data compression and classification model building method. Background technique [0002] Hyperspectral remote sensing is a technology that uses a narrow and continuous spectral channel to continuously image ground objects. In the visible light to short-wave infrared band, its spectral resolution is as high as nanometers, and it usually has the characteristics of multiple bands, with dozens or even hundreds of spectral channels, and the spectral channels are often continuous. Hyperspectral data has gradually become a valuable tool for monitoring the Earth's surface and is used in a wide variety of application scenarios, including agriculture, mineralogy, monitoring, physics, astronomy, and environmental science. Gaofen-5, which was officially put into use in March 2019, is the remote sensing sate...

AI Technical Summary

Problems solved by technology

[0003] Some existing hyperspectral data lossy compression algorithms play an important role in reducing storage device cost and bandwidth, and the decompressed data has good performance in applications such as HS image classification, but the data decompression and feature Re-extraction takes a lot of time, which affects the real-time performance of classification

Since the existing hyperspectral data lossy compression algorithms hardly consider the real-time performance of application scenarios such as classification, it is necessary to find an efficient method to achieve real-time compression and classification of hyperspectral data

Images