A Biologically-Inspired Approach to Making HDR Video Quality Assessment Easier

Quality is easy to recognize by eye. Putting a number on it can be more problematic. For HDR, it is even harder because convenient methods like PSNR and SSIM don't work well. Some newer HDR-aware metrics are now available; however, they can be slow and computationally cumbersome enough to discourage use in real time monitoring. This paper introduces an alternative simple technique intended to be more convenient. Specifically, this paper will present a biologically-inspired method of creating a “Spatial Detail” representation that can be used instead of the original HDR image as a basis for calculating PSNR that has several added advantages. For one, simple correlation analysis of the luma and chroms Spatial Detail signals yields a highly intuitive and sensitive metric of distortion in HDR video. Another advantage is that the Spatial Detail signal provides a guide by which to evaluate distortions of perceptually potent features and highlights separately from distortions of background textures and noise fields. Yet another operational advantage is that the Spatial Detail signal is easy to calculate and thus a candidate for use in real time monitoring.

Published

2016-10

Content type

Original Research

Keywords

High Dynamic Range, Wide Color Gamut, HDR, Spatial Detail, WCG, Ultra HD, UHD, video quality, video distortion

DOI

10.5594/M001679

ISBN

978-1-61482-957-7