Quality Driven Frame Rate Optimization for Rate Constrained Video Encoding

Video quality under rate constraint is mainly controlled by the frame rate and the quantization parameter. This work proposes a mechanism to obtain the optimal frame rate that maximizes video quality under rate constraint. Based on an objective metric of video quality that can reflect subjective quality, this work first proposes a video quality-frame rate-rate constraint model. Second, the relationship between model parameters and video characteristics is formulized. Finally, this work proposes an efficient frame rate optimization mechanism. Experimental results show that the optimal frame rate estimated by our mechanism is identical to the actual optimal frame rate under most bit rate constraints for both training sequences and new test sequences. In addition, the quality loss caused by the estimation error is generally limited within 0.8 dB in our experiments.