A new convex optimization-based two-pass rate control method for object coding in AVS

Abstract

This paper proposed a new convex-optimization-based two-pass rate control method for object coding of China's audio video coding standard (AVS). The algorithm adopts a two-pass methodology to overcome the important interdependency problem between rate control and rate distortion optimization. An exponential model is used to describe the rate-distortion behavior of the codec so as to perform frame-level and object-level rate control under the two-pass framework. Convex programming is utilized to solve for the resultant optimal bit allocation problem. Moreover, the region-of-interest (ROI) functionality is also realized at the object-level. The good performance and effectiveness of this method are illustrated using experimental results.