A real-time merging process for multi-view video coding

摘要

The multi-view video coding (MVC) extension of H.264/AVC is one of the well-known codecs for generating 3D views. It allows for multiple cameras to capture a scene from different views or perspectives to create a 3D output. In this paper, we shall introduce a new method to implement within the codec that enables the merging of multiple matrices of pictures (MOPs) from different participants to produce one output. This proposed technique can be used in several scenarios such as in virtual concerts, where each band member is located at a remote geographical area. At the destination, the received data will be combined to form one output. Additionally, the merging process must be done in real-time in order not to affect the rest of the procedures found in MVC. Three different techniques will be proposed to check the best approach for achieving the required goal. Different criteria or parameters are taken as constraints for the merging process, which are the MOP size, frame dimension, frame rate and number of participants. Experimental results demonstrate a real-time merger from one of the proposed techniques.