Wavelet-based video modelling

Regardless the targeted application (compression, watermaking, texture analysis, indexation, ...), image/video modelling in the DWT domain is generally approached by tests of concordance with some well known pdfs (like Gaussian, generalised Gaussian or Laplace, for instance). Instead of forcing the images/videos to stick to such theoretical models, our study aims at estimating their inner statistical behaviour. In this respect, we first prove that a law modelling the video DWT coefficients decreasingly ordered does exist. Then, we estimate this law by Gaussian mixtures and finally we identify the generality of such model with respect to the data on which it was computed and to the estimation method it relies on. The usefulness of such model was checked-up under the framework of watermarking and indexation applications developed in industrial partnership at the ARTEMIS Department.