EFFECTS OF MEMORY COLOR ON COLOR PERCEPTION OF NATURAL OBJECTS UNDER DIFFERENT ILLUMINATION

Abstract

This study aimed to investigate the effects of memory color on color perception of natural objects under different illuminations. A series of experiments were carried out and launched on the Internet through a website (https://memorycolor.net). Seventy-eight observers participated in the experiments. In Experiment 1, observers were asked to provide a typical color of five natural objects (blueberry, broccoli, salmon, sunflower and tomato) based on their memory. The results showed that memory colors were different for different observers. In Experiment 2, observers were shown an image of natural objects and selected the color they perceived. The experimental images of natural objects consisted of a natural object images taken under daylight simulators D65 and modified images to simulate illumination changes. Four different illuminants were tested: red, green, blue and yellow. The results showed that in almost all cases the perceived colors were closer to the match in D65 than to memory color. This indicated that observers were able to separate the illuminant color from the object and perceived the object color as close to its original image. The contribution of memory color to color constancy was weak. In Experiment 3, Mondrian images having the same colors as the images in Experiment 2 were created to investigate color perception of an image without natural objects. The results showed that the color constancy in the object images was higher than in the Mondrian images. The color constancy index averaged across objects and illuminants for the object image was 68%, and 41% for the Mondrian image. Memory color of the familiar object in the scene was not the only cue for discounting the illuminant.