An Adaptable Low Dimensional Image Generation Model for Eliminating Illumination Influence

HIROSHI, KAMADA and TAKASHI, TORIU and THI, THI ZIN (2017) An Adaptable Low Dimensional Image Generation Model for Eliminating Illumination Influence. In: Sixth International Conference on Advances in Computing, Electronics and Communication - ACEC 2017, 09-10 December, 2017, Rome, Italy.

[img]
Preview
Text
20180215_100758.pdf - Published Version

Download (901kB) | Preview
Official URL: https://www.seekdl.org/conferences/paper/details/9...

Abstract

In general, the color of the image taken by a camera changes when the illumination condition changes. However, human being can perceive almost the true color independent of the illumination (color constancy). The low dimensional image generation model has been discussed as a framework for eliminating illumination influence. It defines how the image color is determined from the illumination color, the object color and the camera spectral sensitivity. Traditionally, the low dimensional image generation model has been identified by setting object color basis functions, illumination color basis functions and the camera spectral sensitivity. In this paper, we propose a method for generating the image generation model by training without specifying any basis functions or camera spectral sensitivity. In that purpose, we prepare some sample images including various colors. Using these data, the image generation model is optimized. The experimental results are shown to confirm the effectiveness of the proposed model.

Item Type: Conference or Workshop Item (Paper)
Uncontrolled Keywords: Color constancy, low dimensional image generation model, object color, illumination color. (key words)
Depositing User: Mr. John Steve
Date Deposited: 10 Mar 2019 09:30
Last Modified: 10 Mar 2019 09:30
URI: http://publications.theired.org/id/eprint/292

Actions (login required)

View Item View Item