Subjective Quality Assessment of Compressed Tone-Mapped High Dynamic Range Videos

TL;DR

This work tackles the challenge of subjective quality assessment for compressed tone-mapped HDR videos by introducing the LIVE-TMHDR database, a large-scale, publicly available resource of 40 HDR sources distorted through 13 TMOs, multiple spatial/temporal parameter settings, and three levels of H.264 compression, totaling 15,000 distorted sequences. The authors conduct a three-phase crowdsourced subjective study, validate reliability via a pilot, and process ratings with SUREAL to obtain high-quality MOS-like scores, while examining inter-subject stability and cross-method consistency. They analyze the impact of TMOs and their parameters on perceived quality and benchmark a broad set of existing and novel objective quality models, finding Cut-FUNQUE and MSML to be the top performers on this dataset. The dataset and findings provide a foundation for improved HDR-to-SDR tone-mapping pipelines and better perceptual quality prediction, with practical implications for streaming HDR content to prevalent SDR displays. The LIVE-TMHDR database is freely accessible at the provided URL and is poised to accelerate research in HDR video delivery and quality assessment.

Abstract

High Dynamic Range (HDR) videos are able to represent wider ranges of contrasts and colors than Standard Dynamic Range (SDR) videos, giving more vivid experiences. Due to this, HDR videos are expected to grow into the dominant video modality of the future. However, HDR videos are incompatible with existing SDR displays, which form the majority of affordable consumer displays on the market. Because of this, HDR videos must be processed by tone-mapping them to reduced bit-depths to service a broad swath of SDR-limited video consumers. Here, we analyze the impact of tone-mapping operators on the visual quality of streaming HDR videos. To this end, we built the first large-scale subjectively annotated open-source database of compressed tone-mapped HDR videos, containing 15,000 tone-mapped sequences derived from 40 unique HDR source contents. The videos in the database were labeled with more than 750,000 subjective quality annotations, collected from more than 1,600 unique human observers. We demonstrate the usefulness of the new subjective database by benchmarking objective models of visual quality on it. We envision that the new LIVE Tone-Mapped HDR (LIVE-TMHDR) database will enable significant progress on HDR video tone mapping and quality assessment in the future. To this end, we make the database freely available to the community at https://live.ece.utexas.edu/research/LIVE_TMHDR/index.html

Figures (13)

• Figure 1: Sample frames from ten HDR source contents, tone-mapped by a professional colorist.
• Figure 2: Convex-hull analysis of low-level descriptors of HDR source videos.
• Figure 3: Distributions of bitrates of compressed tone-mapped videos.
• Figure 4: Example of a structured sampling scheme yielding four representative batches of four videos a piece.
• Figure 5: Distributions of representativeness of 200 batches.