摘要:In this paper, a modified version of the discrete reversible (integer-to-integer) Le Gall wavelet transform (DLGT), distinguishing itself by apparently improved space localization properties and visibly extended potential capabilities, is proposed. The key point of the proposal – ensuring full decorrelation of Le Gall wavelet coefficients across the lower scales. Based on the latter circumstance, a novel exceptionally fast procedure for computing the integer DLGT spectra of the selected image blocks (regions of interest - ROI) is presented. It is shown that the new developments can be efficiently applied to progressive encoding and transmission of image blocks. Progressive encoding and transmission of image blocks is achieved by first transmitting a “rough” estimate of the original image, then sending further details related to one or another image block (ROI). To translate the idea into action, the zero-tree-based encoder SPIHT with an improved quad-tree analysis scheme is employed.DOI: http://dx.doi.org/10.5755/j01.itc.41.3.920
关键词:digital images; discrete wavelet transforms; Le Gall wavelets; Lifting schemes; image processing; progressive image encoding; ROI