Images Subband Approximation in the Task of Bit Representations Volumes Compression

Abstract

Images are an important tool of information processes, as they are a natural form of information exchange for a person, directly perceived by the senses. As a rule, the equipment intended for fixing images contains many recording sensors and a sufficiently large number of binary digits for the digital representation of their values. At the same time, depending on the intended use of visual information, such a volume of data may be excessive and there is a need to reduce it, especially when they are transmitted remotely. This reduction of bit representations of visual information is commonly referred to as compression. The importance of using compression has caused the development of appropriate methods, among which JPEG and JPEG2000 have become the most widespread. Without begging for their merits, we still note their lack of flexibility from the standpoint of the importance of preserving certain properties in the reconstructed approximations of the original images. In this paper, we propose a method of data compression with preservation in the recoverable approximations of spectra in given two-dimensional subdomains of the spatial frequency domain, which is commonly referred to as subband. The results of the conducted comparative computational experiments illustrate the advantages of the proposed subband compression method of images bit representations over the currently most widely used JPEG2000 both in processing time and in the achieved visual quality of the reconstructed approximations.