Demonstration of Color Constancy in Photographs by Two Techniques: Stereoscope and D-up Viewer  Chanprapha PHUANGSUWAN 1 , Mitsuo IKEDA 1 , and Hiroyuki SHINODA 2 1 Color Research Center, Faculty of Mass Communication Technology, Rajamangala University of Technology, Thanyaburi, Pathumthani 12110, Thailand 2 Department of Human and Computer Intelligence, College of Information Science and Engineering, Ritsumeikan University, Nojihigashi, Kusatsu, Shiga 525-8577, Japan (Received September 27, 2013; Revised July 15, 2014; Accepted August 21, 2014) When we look, under daylight, at a scene in a photograph taken under an incandescent lamp, it appears very reddish, showing that color constancy is not maintained. According to the recognized visual space of illumination (RVSI) concept, color constancy should exist in a photograph if one can perceive three dimensions in it. This prediction was confirmed by applying two viewing techniques to perceive a 3D space in a 2D photograph: a stereoscope viewed with two eyes and a D-up viewer viewed with one eye. A wide range of illumination color was investigated, covering range from vivid blue through to vivid orange, and the color constancy index became larger with the 3D perception than with the 2D perception of the photographs produced by the two techniques. # 2014 The Japan Society of Applied Physics Keywords: color constancy, photograph, illumination, stereoscope, recognized visual space of illumination 1. Introduction Color constancy is a well-known and important property of our visual system and many people have studied it and proposed mechanisms to explain it. One explanation is based on the concept of the recognized visual space of illumination (RVSI) developed by Ikeda and co-workers. 1–5) The concept emphasizes the recognition of a 3D space in which objects are placed and the understanding of the illumination filling the space. The visual system then adapts to the illumination and not to the color of objects. 6) The adaptation eventually gives the visual system the property of color constancy for the objects. It is known that color constancy does not normally take place for a photograph. 7) When a photograph is taken in a room illuminated by an incandescent lamp and later viewed in a room illuminated by a daylight type lamp, the photograph appears very reddish, contrary to the color experience in the room where the photograph was taken. Referring to the RVSI theory, Mizokami et al. 8) and Phuangsuwan et al. 9) showed that color constancy exists in a photograph if one can perceive a 3D space in the photograph. To perceive the 3D space they employed a D-up viewer 8,9) with which a subject looked at a photograph with one eye, or monocularly, in the viewer so that the subject viewed only the photograph without seeing any surroundings of the photograph. The brain automatically changed the 2D retinal image of the photograph to a 3D scene, and the subject perceived a white object in the photograph as being almost white in spite of the fact that the object was reddish in the photograph when measured with a color luminometer. Brainard and co-workers insisted that experiments on color constancy be conducted in a real room. 10–13) There are some other ways to perceive a 2D photograph as a 3D scene besides the use of a D-up viewer. The most common technique is to use a stereoscope that employs two photographs taken with the parallax of two eyes, which are viewed binocularly with the respective eyes. In the present study, we employed the stereoscopic technique to confirm color constancy in photographs with 3D perception for a wide range of illumination color from vivid blue through to vivid orange roughly along the Planckian locus. The experiment was carried out at Ritsumeikan University of Japan. In the previous study, 9) which was carried out at Chulalongkorn University of Thailand, only the orange region was investigated. The additional experiment for the blue region was conducted at a later time at Chulalongkorn University to complete the range of illumination; the results will be shown in the present paper in order to confirm the color constancy in photographs with both the stereoscope and the D-up viewer and for both blue and orange regions of illumination. 2. Apparatus Two different sets of apparatus were used, one at Chulalongkorn University (Chula) and the other at Ritsumeikan University (Rits). The former apparatus was used for the orange region of illumination in the previous study 9) and the explanation of the apparatus was already presented in the paper. Here the apparatus was used again for the blue region of illumination by employing a blue film. The apparatus used at Rits is shown in Fig. 1. In the observing room Ro, a subject observed the inside of the room, which was illuminated by a ceiling light of a certain color. Various colors were produced by adjusting the relative intensities of a white fluorescent lamp of three bands type, Lw1, and two fluorescent lamps Lc of the same type as Lw1 but covered with the blue or orange film brought from Chula.  The experiment with the stereoscope was conducted at Ritsumeikan University during the senior author’s 10-month term as a visiting researcher, and the experiment with the D-up viewer was conducted at Chulalongkorn University when she was a PhD student. OPTICAL REVIEW Vol. 21, No. 6 (2014) 810–815 810