Multiscale-based image enhancement

Abstract

Wavelet-based image analysis offers the opportunity to enhance images using features extracted at different scales and subbands. We propose a two-part wavelet domain image enhancement method. First, we apply a locally adaptive filter to the wavelet transform detail coefficients to simultaneously suppress noise and enhance edge contrast. Our estimates of noise variance behaviour across the decomposition scales and subbands are based on simulations. To enhance contrast between large, flat image regions, we apply global histogram equalisation to the wavelet transform approximation coefficients at the coarsest decomposition level.