Rate-Prediction Structure Complexity Analysis For Multi-View Video Coding Using Hybrid Genetic Algorithms

Yebin Liu,Qionghai Dai,Zhixiang You,Wenli Xu

DOI: https://doi.org/10.1117/12.703849

2007-01-01

Abstract:Efficient exploitation of the temporal and inter-view correlation is critical to multi-view video coding (MVC), and the key to it relies on the design of prediction chain structure according to the various pattern of correlations. In this paper, we propose a novel prediction structure model to design optimal MVC coding schemes along with tradeoff analysis in depth between compression efficiency and prediction structure complexity for certain standard functionalities. Focusing on the representation of the entire set of possible chain structures rather than certain typical ones, the proposed model can given efficient MVC schemes that adaptively vary with the requirements of structure complexity and video source characteristics (the number of views, the degrees of temporal and interview correlations). To handle large scale problem in model optimization, we deploy a hybrid genetic algorithm which yields satisfactory results shown in the simulations.