Method of lossless image compression in the Earth remote sensing systems

Abstract

A method of lossless digital image compression is presented in the paper. Its implementation demands minimum computing resources and it can be efficiently used in Earth remote sensing systems. The method makes possible maximum removal of statistical redundancy of images without spectral transformations. A digital image is resolved into bit planes which are processed independently of one another. That allows their parallel processing without any restriction on the pixel length of the image. Each bit plane is approximated by a two-dimensional Markov random field with discrete values. Prediction of elements of the bit plane is accomplished on the basis of the theory of conventional Markov processes. Only operations of logical comparison are used for the prediction, whereas computational operations such as multiplication and addition are not required, which makes it possible to increase the speed of image processing. The research conducted shows that the method is not inferior to known methods of lossless compression in its efficiency, while being considerably superior to them in the speed of processing. The method is applicable both to panchromatic and color images and can be used in any systems of digital image processing.