Multi-feature 360 Video Quality Estimation

TL;DR

This paper introduces a multi-feature quality assessment method for 360-degree videos that combines various objective features on viewports to better predict perceived quality, outperforming existing metrics.

Contribution

It presents a novel viewport-based quality metric that integrates multiple features through a learned model, improving accuracy over state-of-the-art methods.

Findings

Outperforms existing 360-degree video quality metrics

Effective across different projection methods and distortion types

Validated on large datasets with cross-dataset testing

Abstract

We propose a new method for the visual quality assessment of 360-degree (omnidirectional) videos. The proposed method is based on computing multiple spatio-temporal objective quality features on viewports extracted from 360-degree videos. A new model is learnt to properly combine these features into a metric that closely matches subjective quality scores. The main motivations for the proposed approach are that: 1) quality metrics computed on viewports better captures the user experience than metrics computed on the projection domain; 2) the use of viewports easily supports different projection methods being used in current 360-degree video systems; and 3) no individual objective image quality metric always performs the best for all types of visual distortions, while a learned combination of them is able to adapt to different conditions. Experimental results, based on both the largest available 360-degree videos quality dataset and a cross-dataset validation, demonstrate that the proposed metric outperforms state-of-the-art 360-degree and 2D video quality metrics.