Multi-view video prediction architecture using faster prediction order

Hardware architecture of multi-view video systems requires more power consumption and chip area than single-view and stereo-view systems because the number of channels for encoding views is increased linearly depending on the number of cameras. Also, encoding time between channels is different and channels with inter- and inter-view prediction take more encoding time than other channels that require only intra- or inter prediction. In this paper, we propose a multi-view video prediction architecture using a novel prediction order for reducing the encoding time of MVC, improving hardware performance and solving increased processing time gap between channels. The experimental results show that the proposed scheme can reduce the power consumption, chip-area and processing time during the MVC encoding by 22.81% and 23.64% and 8.58% respectively compared with other schemes.