A novel method of adaptive GOP structure base on coding efficiency and complexity joint model

Abstract

In this paper, we focus on the models that the coding time and efficiency affected by GOPsize, and then find that large GOPsize will usually bring the more coding time and less coding efficiency, and vice verse. We analyze the reason which resulting in the above circumstances and propose two models: coding efficiency and time. Their maximum errors are less than 5% and 2.7%, respectively. In practice, the coding system is expected to reduce coding time while improve coding efficiency as possible. Base on the two models, we further propose a joint function of choose GOPsize adaptively to achieve the optimal tradeoff between coding time and efficiency. We improve the joint function to make it adapt to different system requirements by changing its parameters.