Frame Rate Optimization of Video Encoding Using Subjective Quality Metric

ABSTRACT

With the improvements of video coding technology, network infrastructures, storage capacity, and CPU computing capability, the applications of multimedia systems become wider and more popular. Nowadays, how to provide the best video quality to users under the rate constraints is always an important issue in video coding. In general, we can either increase the quantization step size or reduce the frame rate to meet the bitrate constraint. Objective quality metrics such as PSNR has long been used as the quality assessment in video coding. However, in the condition of video frame rate switching, this objective quality metric is not able to reveal the perception of human eyes. Instead, a good subjective video quality metric is necessary in helping us find the best encoding configurations.

In this work we focus on how to find the optimum frame rates in the sense of maximizing a subjective video quality metric under different rate constraints. We first characterize video sequences by parameters in temporal and spatial aspects such as edge strength, average motion vector, and motion compensation difference. Then the subjective video quality metric of a video sequence is modeled by these characteristic parameters. Finally, with this model a frame rate optimization algorithm in video encoding is proposed.

Simulation results show that reducing frame rate may be more effective than increasing quantization step size when the given bitrate is not sufficiently high. It also reveals that the frame rate provided by the proposed algorithm is very close to the ideal case.