Investigating subjective attributes of quality for videos displayed with local backlight dimming

This paper investigates what composes the quality of videos displayed on LCD with local backlight dimming. In a subjective experiment, participants assessed the level of nine attributes defined using the Qualitative Descriptive Analysis method. Results show that three attributes (Contrast, Change over time and Grain) vary significantly depending on the backlight dimming algorithm. Predicting global quality grades from the attributes using Partial Least Square (PLS) regression does not provide satisfying results. However, the attributes can be represented through objective measures and a similar model mixing attributes and one objective measure can predict global quality satisfyingly.