.png){kind=icon}
.png){kind=icon}
Return
Compression Efficiency and Picture Quality Assessment of Broadcast HDR Videos With and Without Film-Grain
.png){kind=icon}
PRE
AI
G
J
A
DOI：10.1109/TBC.2026.3651199.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
Along with the ever-growing popularity of information rich video formats such as High Dynamic Range (HDR), Wide Colour Gamut (WCG) and Ultra High Definition (UHD), the need for efficient distribution of such video content has become crucially important. To this end, video compression plays an important role in reducing the bitrate requirements by, for example, up to two orders of magnitude. This work first investigates the performance of state-of-the-art video compression standards with the help of a large-scale subjective study. The study for the first time provides some useful insight into the types of artifacts that are introduced by the video codecs on HDR content with film-grain. The evaluation of the pictures generated by the video codecs is also a crucial aspect in the video distribution pipeline. For this purpose, this paper further provides an HDR picture quality metric which is based on an existing Standard Dynamic Range (SDR) picture quality metric called Detail Loss Metric (DLM). The proposed metric includes several novel features to make it suitable for assessing compression artifacts of HDR content with and without film-grain. They are 1) A Just Noticeable Difference (JND) based perceptual weighting function that captures the non-linearity in the HDR signal. 2) An entropy masking module to capture the impact of film-grain characteristics. 3) A contrast masking operation based on previous work, to better represent the impact of local visual masking. Performance of the proposed method and several state-of-the-art HDR and SDR quality metrics is evaluated on two large-scale datasets using three commonly used accuracy measures: Pearson Linear Correlation Coefficient (PLCC), Spearman’s Rank Correlation Coefficient (SRCC) and Root Mean Square Error (RMSE). A further measure of how reliably a given metric can represent the user perception is also introduced. The experimental results indicate superior performance of the proposed method, demonstrating a PLCC score of more than 94% and similar good performance scores for other accuracy measures.
Keywords:
Video quality metrics
video codecs
film-grain
HDR
Journal
IF:
4.8
Papers:
2.1K
Citations:
3.0K
Organization
Cited Papers