An adaptive discrete wavelet transforms method of grayscale images

Abstract EN

In this paper, a modified discrete wavelet transform method with a hierarchical adaptation to the signal size is developed to enhance wavelet-compression performance: reduction of the bit rate, enhancing the quality of reconstruction, and the reduction of computational complexity.

The method is based on adaptive padding alignment wavelet-coefficients to an approximation coefficient sequence of a certain wavelet-decomposition level for forming an even length sequence.

The presence or absence of the alignment wavelet-coefficients is determined by the binary values of the alignment vector components.

A comparative efficiency evaluation, of the methods of onedimensional and two-dimensional rational (or integer) wavelet-transform with and without a hierarchical adaptation to the image size, is presented.

It is found that the developed method provides a gain in terms of compression coefficients, peak signal-to-noise ratio and computational complexity.