Abstract
Recent remarkable advances in display technology enable us to present visual stimuli with higher color depths. For instance, the current standard middle-range graphic boards and displays support 10-or-more-bit color depth for each of RGB channels. With those high color depth display systems, the standard gamma correction procedure to linearize the relationship between video inputs and luminance outputs may not be efficient in the following two reasons. First, the input/output relationship of the high color depth systems can be calibrated more accurately within a limited local color space since we can collect sufficient number of luminance/color samples around the target range, while the standard gamma correction procedure requires sparse measurements in a whole range of colors to describe the relationship with an exponential model. Second, a color transformation matrix obtained from the maximum RGB inputs to produce the required chromaticity may not be ideal in characterizing high color depth displays due to the nonlinearity of the systems. Therefore, we propose a novel display characterization procedure for high color depth displays. In this procedure, all the calibrations and estimations can be achieved only in a local luminance/color space, ignoring the irrelevant input range. Our aim is especially focused on developing a fairly quick and efficient method for finding display inputs that produce specific pre-specified luminance and chromaticity outputs. Specifically, our method consists of two estimation steps. First, the linearity between video inputs and luminance outputs is attained by measuring luminance outputs in a limited input space and by generating color lookup tables for that local space (Local Gamma-Correction). Second, the required RGB video input values are assessed by a local color transformation matrix estimated by least-squares estimations (Local Color Estimation). The whole procedures are integrated into GUI-based display characterization software, Mcalibrator2, written in MATLAB. The software is publicly available (https://github.com/hiroshiban/Mcalibrator2).
Meeting abstract presented at VSS 2016