Modeling LCD Displays with Local Backlight Dimming for Image Quality Assessment

Traditionally, algorithm-based (objective) image and video quality assessment methods operate with the numerical presentation of the signal, and they do not take the characteristics of the actual output device into account. This is a reasonable approach, when quality assessment is needed for evaluating the signal quality distortion related directly to digital signal processing, such as compression. However, the physical characteristics of the display device also pose a significant impact on the overall perception. In order to facilitate image quality assessment on modern liquid crystal displays (LCD) using light emitting diode (LED) backlight with local dimming, we present the essential considerations and guidelines for modeling the characteristics of displays with high dynamic range (HDR) and locally adjustable backlight segments. The representation of the image generated by the model can be assessed using the traditional objective metrics, and therefore the proposed approach is useful for assessing the performance of different backlight dimming algorithms in terms of resulting quality and power consumption in a simulated environment. We have implemented the proposed model in C++ and compared the visual results produced by the model against respective images displayed on a real display with locally controlled backlight units.