A general effective rate control system based on matching measurement and inter-quantizer

Abstract

Current rate control systems use the frame complexity to determine the frame-level bitrates. These methods ignore the dependence between the current frame and its reference frames, degrade the accuracy of their rate distortion models, and reduce the rate control performance as a consequence. In this paper, we propose a general effective rate control system based on matching measurement and inter-ρ quantizer (RCMMAIL) for actual video, which includes two phrases: (1) a novel Frame-Level Rate Control algorithm based on Matching Measurement (FRCMM) is proposed to compute the frame-level target bitrates, which measures the frame matching score to characterize the dependency between the current frame and a reference frame. With this quantified dependency, a more accurate rate distortion model is obtained; (2) based on the frame-level target bitrates, a novel inter- ρ quantizer is proposed to compute the corresponding quantization parameters for each frame. This quantizer can solve the problem of the uncertain parameters in the current ρ model, improving the accuracy and reducing computational complexity simultaneously. Experimental results show that the proposed rate control system achieved better Peak Signal-to-Noise Ratio (PSNR) than other compared models. Meanwhile, the average relative rate control error and time consumption is reduced.