Distributed video coding of multiview sequences

Since 2002, distributed video coding has become a major paradigm, because of its attractive theoretical results, and its promising target applications. Indeed, in such a compression system, all inter frame comparison is shifted from the encoder to the decoder, which im- plies an important complexity reduction at the encoder, and moreover, an independent encoding of the camera in case of multiview compression. This thesis aims at proposing new solutions for distributed video coding, and especially within multi-camera setting. These contributions handle several aspects of distributed video coding paradigm as: a new rate-distortion model and its applications, novel side information generation techniques and finally a detailed study of the Wyner-Ziv decoder. All these new approaches aim at enhancing the rate-distortion performance or at leading to a better comprehension of the coder behavior. These ones are explained in detail in this manuscript preceded by a complete overview of their context.