Regression Approach to Lossles Compression Algorithm for Hyperspectral Images

At present, there is a significant increase in interest in solving applied problems using hyperspectral aerospace images obtained from satellites of spacecraft for remote sensing of the Earth. Hyperspectral images show significant spectral correlation, the use of which is critical for compression. In this paper, we propose an efficient approach to hyperspectral image compression using a lossless regression algorithm. The main idea of the proposed transformation is an algorithm with finding pairs of correlated channels and then creating lossless transformed blocks using regression analysis, which makes it possible to reduce the size of the aerospace image channels and transform them before compression.