Dependent bit allocation in multiview video coding

We consider the bit allocation problem in multiview video coding (MVC). A dependent coding technique using trellis expansion and the Viterbi algorithm (VA) is proposed, which takes into account dependencies across time and views. We note that, typically, optimal quantizer choices have the following properties: i) quantization choices tend to be similar for frames that are consecutive (in time or in view), ii) better quantization tends to be used for frames closer to the root of the dependency tree. We propose a search algorithm to speed up the optimization of quantization choices. Our results indicate significant gains can be achieved by an appropriate selection of bit allocation across frames.