An HTTP Adaptive Streaming Method Considering Motion Intensity

Abstract

This paper proposes an HTTP adaptive streaming method considering motion intensity of video scenes for improving quality of experience (QoE) performance. Motion score is used as one of the metrics for video multimethod assessment fusion (VMAF), which represents motion intensity. The proposed method calculates the motion score of each two-frame difference in advance. Based on the motion scores, a client adaptively selects the quality of the next video segment. The proposed method selects a high-quality segment when the motion intensity of the scene is high in order to improve QoE performance, whereas in order to prevent stalling, the proposed method selects a low-quality segment for a low-motion segment. We implement the proposed method on a real system to evaluate the performance, and show the effectiveness of the proposed method.