This study aims at developing an image quality metric for camera auto white balance (AWB), with a transform to just noticeable differences (JNDs) of quality in pictorial scenes. In this study, a simulation pipeline was developed for a Nikon D40 DSLR camera, from raw capture to rendered image for display. Seven real-world scenes were used in the study, representing capture conditions in outdoor daylight, indoor fluorescent lighting, and indoor incandescent lighting conditions. Two psychophysical experiments were performed, and 38 observers participated in the study. In study one, method of adjustment was used to explore the color aims for individual scenes. In study two, a softcopy quality ruler method was used to refine the color aims and define the quality falloff functions. A quartic function was used to fit the results from the softcopy ruler study, forming the proposed objective metric for camera auto white balance.
Elaine W. Jin, Yixuan Wang, Wentao Liu, "Development of a Perceptually Calibrated Objective Metric for Auto White Balance" in Proc. IS&T Int’l. Symp. on Electronic Imaging: Image Quality and System Performance XV, 2018, pp 342-1 - 342-6, https://doi.org/10.2352/ISSN.2470-1173.2018.12.IQSP-342