An alternative multiscale framework for variational perceptually-inspired contrast enhancement of color images
Baptiste Mazin, Edoardo Provenzi · Color and Imaging Conference · 2018
We present a general multiscale strategy for perceptually-inspired contrast enhancement of color images. The idea behind this methodology comes from a recent wavelet-based variational framework for contrast intensification. We will show that the equations for the wavelet coefficients coming from the variational setting can be re-written in a more general multi-resolution framework, where the only requirement is the existence of an approximation and a detail layer at each different scale. In particular, we will show that a Laplacian pyramid implementation of the variational algorithm performs faster and better than the wavelet-based one. These results open the possibility to efficiently apply the contrast enhancement equations also to video sequences.