Anaglyph image generation by matching color appearance attributes

• The paper proposes an anaglyph image generation algorithm.
• The algorithm aims to reproduce accurate perceptual color appearance attributes.
• Subjective tests are conducted to compare four anaglyph generation methods.

Anaglyph is an inexpensive three-dimensional (3D) displaying technique. It has deficiencies like color distortions, retinal rivalry, ghosting effect, etc. In this paper, a new method is proposed for anaglyph image generation. In contrast to most existing studies which focus on solving part of the problems, the proposed method tries to minimize the color distortions, retinal rivalry, and ghosting effect simultaneously. It works in the CIELAB color space which is perceptually uniform. Rather than matching the L⁎a⁎b⁎L⁎a⁎b⁎ values of the stereo image pair and the perceived anaglyph image as in an existing method (McAllister et al., 2010 [1]), the proposed algorithm is aimed at matching perceptual color appearance attributes which can accurately define the color perception of the human visual system (HVS). Rather than evaluating the performance based on typically 3–5 images as in most prior studies, subjective tests have been conducted, involving 25 stereo image pairs and 20 subjects, to compare four anaglyph generation methods. The proposed method demonstrates a very good performance for most of the images in the subjective test.