In this article, the rate-distortion performance for coding depth maps of the new 3D video format using the multiscale pattern matching algorithm, MMP (Multidimensional Multiscale Parser), is presented. In addition, an hybrid architecture for multiview image plus depth compression is proposed. MMP’s coding efficiency contributes to the preservation of high frequency details of an image, which in the case of depth maps represent the objects’ edges in a scene. Hence, the reconstructed views present well defined edges, which are an important characteristic for view synthesis algorithms. We apply the H.264/MVC for coding the multiple views, and MMP for coding the respective depth maps. With this new proposal, the reconstructed views present better defined borders and objective quality gains of up to 1 dB, if we compare our results with an architecture based only on the H.264/MVC.