Color Restoration Algorithm Based on Color Instance Under Low Illumination

Abstract

Color restoration algorithm using color instance-based reasoning is proposed. The proposed system does not employ a physical model but color instance. The color instance, Color Change Vector, is defined based on distribution of color values for 175 color scheme cards taken under low and standard illuminations. Modification of CCV is proposed based on color value and linguistic expression. The given image is restored by adding the appropriate color instances for a given color value. A prototype system for color restoration is constructed. An experiment is done with dynamic images of an outdoor walking person, to evaluate the performance of the proposed system in terms of color-difference and processing time. The proposed method presents a foundation for identifying a person captured by a practical security system using a low cost CCD camera.