Postprocessing of low bit-rate wavelet-based image coding using multiscale edge characterization

In this paper, we propose a new postprocessing method for low bit-rate wavelet-based image coding which uses the technique of wavelet modulus maximum representation (WMMR). The edge degradation from wavelet-based coding is discussed under the overcomplete wavelet expansion, and interpreted as the distortion of wavelet modulus maxima, i.e., magnitude decays. Based on the empirical analysis and experimental results, we develop a set of compensation functions to restore the distorted wavelet modulus maxima of a coded image. Therefore, we can enhance the image quality for low bit-rate wavelet coding by reconstructing the coded image using the WMMR with improved visual quality and image fidelity (peak signal-to-noise ratio).