Using binocular energy modeling for stereoscopic color image coding

In this paper, we propose a coder for stereoscopic color images based on the binocular properties of the human visual system (HVS). In the preprocessing stage we modeled some properties of the simple and complex cells. These cells characterized by their orientation and amplitude are responsible for binocular fusion; they take as input a set of signals representing the two retinal images and give as output binocular signals. To model this process we have used mathematical functions that have the same characteristics as the simple and complex cells such as wavelet and bandelet. After the matching process, we obtain a residual image, a disparity map and the reference image; these allow to predict the target image. The residual image contains the matching error. The one obtained with our approach contains a very few amount of data generating low bitrate. The experimentation stage showed that our coder gives better results than the two famous coders coming from literature.