Efficiency Of Lossy Compression Of Noisy And Pre-Filtered Remote Sensing Images

This paper state that pre-filtering of noisy image before compression can provide certain benefits only under several conditions. First, a noisy image to be compressed has to have a rather simple structure. Second, noise intensity is to be rather high; only in this case pre-filtering is able to produce considerable improvement of visual quality compared to noisy image. In the case when a pre-filtered image is subject to compression, benefits due to pre-filtering are observed only if compression ratio (quantization step) is not too large. Moreover, we have not observed OOP for pre-filtered and then compressed images. Thus, their quality (according to both standard and HVS-metrics) permanently decrease if compression ratio increases. If CR has to be large, there is no reason to perform pre-filtering of noisy images.