AN APPROACH TO PREDICTION AND PROVIDING OF COMPRESSION RATIO FOR DCT-BASED CODER APPLIED TO MULTICHANNEL REMOTE SENSING DATA

摘要

We propose and study an approach to compression ratio prediction and providing with application to multichannel remote sensing images. DCT-based compression techniques, both 2D applied component-wise and 3D applied to sub-band groups, are considered. It is shown that compression ratio can be predicted for component-wise processing of multichannel images using a simple modification of approach proposed recently for coding single-component images, either noise-free or corrupted by different types of noise where an input parameter used in prediction is percentage of zeros in AC DCT-coefficients after quantization. It is also demonstrated that compression ratio can be predicted for 3D coders applied to multichannel image as a whole or as a set of groups of channels. Such a prediction, under certain conditions, allows also providing a desired compression ratio within a procedure that does not require multiple compression. The approaches are tested for Landsat multichannel remote sensing data.