Motion Based Reference-Free Quality Estimation for H.264/AVC Video Streaming

Abstract

The scope of this paper is the estimation of subjective video quality for low resolution video sequences as they are typical for mobile video streaming. Although the video quality experienced by users depends on spatial (edges, colors, ...) and more considerably on temporal (movement speed, direction, ...) features of the video sequence, most of the proposed methods are based on spatial features. This paper presents a new reference free approach for quality estimation based on motion character. The character of motion is determined by the amount and direction of the motion between two scene changes. In this paper, the design of a universal reference-free video quality metric is presented. The design of the quality metric is based on content adaptive parameters, allowing for content dependent video quality estimation. The performance of the proposed universal metric was evaluated with various video content types and compressions settings. The results show that the proposed approach provides powerful means of estimating the video quality experienced by users for low resolution video streaming services