A Constrained Spatial-Temporal Frame Rate Control Model of Conversational Video for Multipath Transport System

Conversational video service is becoming increasingly popular on the Internet. However, higher video resolution and frame rate are often accompanied by enlarging the bit rate of the video coding output, and further lead to a large bandwidth requirement. In this paper, we propose a constrained spatial-temporal frame rate control model for conversational video service, which combines video coding with the feedback of data transmission in multipath transmission to enhance rate control. The goal of the model is to optimize the constraint relationship among frame rate, resolution and bitrate, and achieve a better QoE of conversational video service. The target bitrate and frame rate are determined on the fly by considering the network condition. Moreover, the model introduces a new concept termed as region rate, which allows different regions (region of interests and non-region of interests) of a frame to have different rates in accordance with their importance. Furthermore, the paper adopts the perceptual video compression method to improve the perceptual quality, which allocates more resources to the ROIs and selects paths with appropriate status to transmit ROIs data. The experimental results show that the proposed method could control the bit rate while ensuring the video quality and realize the joint optimization of frame rate, resolution and bit rate in comparison to existing approaches.