Color Instance-Based Reasoning and its Application to Dynamic Image Restoration Under Low Luminance Conditions

Abstract

Color instance-based reasoning is applied to dynamic image restoration under low luminance conditions is presented by introducing \textit{color change vectors}. The color change vectors are constructed based on image measurement, i.e., no optical
lighting models are required. This property enables us to deal with low luminance conditions, under which it is difficult to capture color information using traditional optics-based methods. The proposed algorithm restores color values by adaptively modifying the color change vectors in response to a given dynamic image. Experiments are done with still and dynamic images to compare the proposed algorithm with the conventional one in terms of color difference. The experimental results show that the proposed algorithm decreases the color-difference a maximum of 20% compared to the conventional algorithm. The proposed algorithm presents the foundation to identify a person by a low cost CCD camera in the practical security system.