Distributed video coding is a new video coding paradigm that allows exploiting the source statistics at the decoder only. In this scenario, two correlated sources are independently encoded using separated encoders and the bit streams associated to each one are jointly decoded exploiting the correlation between them. Improved error resilience is another major functionality of this new video coding paradigm since the usual encoder prediction loop and the associated error propagation do not exist anymore, because the signal correlation is not explored in the encoder. Therefore, this thesis aims to present a new architecture for distributed video coding, analyzing and developing more efficient algorithms for it, thus reducing the gap in performance when it is compared to traditional video coding [68]. Therefore, the codec proposed in this thesis provided: a DVC architecture optimization; the development of new tools for side information generation process; inclusion of a module for measuring correlation between frames in order to assist in the Wyner-Ziv frame reconstruction; the use of a Wyner-Ziv variable quantization process; the option of partial coding of bitplanes; a correlation channel model based on the estimation of parameters in band and DCT coefficient levels; an inclusion of a normalization stage to the side information; and the development of a channel code more appropriate to the proposed architecture.