Pre-requisites for smart lossy compression of noisy remote sensing images

Remote sensing images are usually subject to compression for their further transmission, storage and dissemination. Because of lossy nature of compression, resulting images appear distorted. Degradations of image quality due to compression depend on noisy input image, a type and intensity of noise, and used image coder. To control image degradations, for a given coder, one should predict compression performance to be able to properly choose coder parameter(s). In this paper, we present pre-requisites for such a controlled lossy compression of noisy remote sensing images. The main attention is paid to image coders which are based on discrete cosine transform, due to relatively simple adaptation of its main parameter, quantization step, for controlling the effect of compression.