Towards acomprehensive color stereoscopic matching model based on HVS behavior

3D applications need a stereo matching algorithm that can merge the left image with the right image. In this paper we propose a stereo matching algorithm based on some binocular properties of the human visual system. For this, we have focused on the simple and complex cells properties that are responsible for binocular fusion. A model, based on a complex wavelet transform, a color contrast sensitivity function and a bandelet transform, is proposed to model the properties of the simple and complex cells characterized by their size, orientation, phase and amplitude. This model is designed to calculate the binocular energy generated by each binocular complex cell. A model refinement based on properties specific to the lateral geniculate body is then proposed to refine the results obtained with the binocular energy model calculation. The obtained results can be exploited by the stereoscopic vision applications, such as 3D reconstruction, stereoscopic coding, stereoscopic quality assessment and other applications.