<img height="1" width="1" src="https://www.facebook.com/tr?id=414634002484912&amp;ev=PageView &amp;noscript=1">
Donate
Winter Boot Camp for IP Networking Professionals
Donate

A New Approach for Measuring Perceived Brightness in HDR Displays

August 18, 2021

Researchers at Samsung have developed a new technique for measuring and predicting perceived brightness of High-Dynamic Range material by harnessing the Helmholtz-Kohlrausch (H-K) effect, where the intense saturation of spectral hues (notably reds and blues) can increase the color’s perceived luminance. As described in their paper in the August SMPTE Motion Imaging Journal, “A Visual Model for Very Wide-Gamut HDR Displays That Accounts for the Helmholtz– Kohlrausch Effect,”https://www.smpte.org/motion-imaging-journal this methodology could be used by display manufacturers to increase perceived brightness without increasing power (and therefore cost) in displays with wide-gamut and spectrally pure primaries. HDR televisions and computer monitors with the same amount of luminance and a greater color range can thus appear brighter, more vivid and potentially more natural to viewers.

In their experiments, the team at the Samsung America Display Lab built upon and extended the iCAM06 image color appearance model originally designed for HDR image evaluation, incorporating an improved treatment of these vivid colors especially affected by the H-K effect. They chose their test video sources to be typical of cinematic content rather than strongly saturated advertisements. They separated frequency, color and spatial image data, and then processed each to extract the appropriate image information.

  • As in the traditional iCAM06 model, they filtered the spatial frequency data into a baseline layer and a detail layer of high-frequency edges.
  • In the first enhancement to the iCAM06 model, they used color masks to identify color data clusters, usually areas of saturated reds and blues where the H-K effect is particularly strong, and then calculate these H-K effects.
  • The second enhancement used masks to identify spatial clusters and build highlight objects. Because the highlight object layer in their test images usually -- but not always -- corresponded to the brightest portions of the image, the team opted to perform the spatial masking manually, though they note the potential to automate this step in the future using machine learning.

In their model flow, the baseline layer, color data clusters and spatial clusters then underwent color adaptation and tone compression corrections, with added modification to correct the H-K effects on the color data. With the addition of the detail layer, they calculated the average luminance. In order to compare this calculated luminance with the viewer’s perceived brightness, they needed to include Human Factors corrections. However, human vision works differently from a computational model, and it’s not possible to switch H-K corrections off or on in the brain. So the team devised trials to simulate both week and strong H-K effects. Their tests included subjective human vision trials with viewers having 20/20 or corrected 20/20 vision, and no color impairment.

Their results confirm that a wider color gamut and greater dynamic range drives viewers to perceive increased brightness, and that the viewers overwhelming choose the displays with more saturated, spectrally pure color. The team found that, “Our model extension of iCAM06 predicts the H-K effect well. The model can predict increased brightness from a display based on its wider colorimetry when compared to a display with equal luminance but less saturated colors. It can predict with certainty that viewers are sensitive to the H-K effects, and that increasing contrast in red or blue leads to the perception of a brighter display.”

Dig deeper into Samsung’s computational and test models for predicting perceived brightness of HDR display devices. Read the complete article in the SMPTE Motion Imaging Journal. https://ieeexplore.ieee.org/document/9508135

Tag(s): Featured , HDR , News

SMPTE Content

Related Posts