Projecte final de carrera realitzat en col.laboració amb University of Southern California

This thesis presents a new wavelet transform speci cally designed for the coding of depth images which are used in view synthesis operations. Two basic properties of these images can be leveraged: rst, errors in pixels located near the edges of objects have a greater perceptual impact on the synthesized view; second, they can be approximated as piece-wise planar signals. We make use of these facts to de ne a discrete wavelet transform using lifting that avoids ltering across edges. The lters are designed to t the planar shape of the signal. This leads to an e cient representation of the image while preserving the sharpness of the edges. By preserving the edge information, we are able to improve the quality of the synthesized views, as compared to existing methods. i