Goal-seeking approaches to characterize non-CRT as well as CRT displays for vision experiments

摘要

Display characterization is an important part of the experimental procedures in vision science since almost all current experiments are controlled by computers and the visual stimuli are presented on computer displays. A two-step procedure (gamma-correction and the following linear color transformation) has been widely used to calibrate display luminance and chromaticity profiles to ensure that visual stimuli are presented accurately. However, the standard procedure is not always valid to characterize recent new types of displays such as LCD (Liquid Crystal Display), DLP (Digital Light Processing), and OLED (Organic Electro-Luminescence Display) because the method was developed based on the internal model of CRTs (Cathode Ray Tube). Furthermore, these new types of displays have come into the mainstream of display products and it becomes increasingly difficult to obtain CRTs. Therefore, vision researchers are required to develop new display characterization approaches that are appropriate to any display type. Here we propose novel display characterization methods that are applicable to non-CRT as well as CRT displays. Our aim is especially focused on developing fairly quick and efficient methods for finding display inputs that produce specific pre-specified luminance and chromaticity output values. Our new methods use a data-driven gamma-correction procedure combined with recursive-linear and non-linear goal-seeking algorithms. The methods are relatively model-free, only assuming a piece wise linearity of the system in the initial estimation step. The whole procedures are integrated into GUI-based display characterization software, Mcalibrator2, written in MATLAB (Ban and Yamamoto, J. Vis. (2013). The software is now publicly available and can be downloaded from our website). The present study further tested the efficiencies and accuracies of our new methods to produce chromatic visual stimuli on non-CRT displays. The results showed that our methods significantly outperformed the standard procedure and were applicable to a wide range of display devices.