Affiliations: Department of Applied Mathematics, Kaunas University of Technology, Studentų 50, LT-51368 Kaunas, Lithuania, e-mail: jonas.valantinas@ktu.lt, deividas.kancelkis@stud.ktu.lt, gintare.visciute@ktu.lt | IT Department, Money Supermarket House, St. David's Park, Chester (Ewloe), CH53UZ, United Kingdom, e-mail: rokas.valantinas@gmail.com
Note: [] Corresponding author.
Abstract: In this paper, a modified version of the discrete wavelet transform (DWT), distinguishing itself with visibly improved space localization properties and noticeably extended potential capabilities, is proposed. The key point of this proposal is the full decorrelation of wavelet coefficients across the lower scales. This proposal can be applied to any DWT of higher orders (Le Gall, Daubechies D4, CDF 9/7, etc.). To open up new areas of practical applicability of the modified DWT, a novel exceptionally fast algorithm for computing the DWT spectra of the selected signal (image) blocks is presented. In parallel, some considerations and experimental results concerning the energy compaction property of the modified DWT are discussed.
