A Novel Meteosat Second Generation Image Compression Method Based on Radon Transform, Linear Predictive Coding with Filtering and Sorted Run Length Coding

Abstract

The monitoring of the earth by meteorological satellites became an indispensable process. The main mission of these satellites consists to improve daily weather forecasts and anticipate severe weather. But they also follow a few activities such as: the monitoring of urban spaces, the evolution of the vegetation cover, and the detection of forest fires etc. The Meteosat Second Generation satellite is equipped with a radiometric sensor SEVIRI (Spinning Enhanced Visible & Infrared Imager), allows the acquisition of 12 image files that correspond to the 12 channels of visible and mid-infrared, thermal infrared every 15 minutes. The accumulation of these images confront us with problems of storage and transmission. There are many interesting techniques of image compression that allow an efficient minimization of the size of the images with a good rate-distortion compromise. This paper presents a novel Meteosat Second Generation image compression method based on Radon transform and two novel techniques, which are Linear Predictive Coding (LPC) with filtering and the Sorted Run Length Coding (SRLC). The Radon transform is applied to MSG images to retrieve the Radon points. The residual predictions are extracted first from the filtered Radon points and then encoded by the arithmetic encoding to recover a single codeword using the LPC coding with filtering. Finally, the SRLC coding is performed to achieve a high and fixe compression ratio by the generation of a sorted sequence of the codeword and a sorted occurrence table. This method ensures an excellent rate-distortion compromise: high and fixed compression ratio of 99.90 % while preserving the integrity and the quality of the MSG images after decompression.