Considering Chromatic Adaptation in Camera White Balance

White balance is an important process in image signal processing (ISP) to adjust the overall tone of an image, which actually simulates the chromatic adaptation mechanism in the human visual system. A typical white balance always aims to completely remove the color cast of the illumination, so that the color appearance of the image appears as it is captured under daylight (i.e., D65 illumination). This assumes the human visual system can always achieve a complete chromatic adaptation under any illumination. In this study, we built a scene under 85 different light settings having a wide range of chromaticities. The human observers viewed the scene under each light setting, and then adjusted the color appearance of the scene image on a smartphone display to match that of the scene. A pipeline was designed to apply a chromatic adaptation transform to an image, which changes the color appearance of the image by changing its white point. The results clearly showed that it is not appropriate to always adjust the image white point to D65. Therefore, white balance should consider the effect of the scene illumination on the degree of chromatic adaptation, and adjust the image white point accordingly.