Robust color object tracking method against illumination color Change

Abstract

Color information is one of the most important keys in object tracking. Color appearance of objects depends on illumination color. Therefore, object tracking by color information does not work well in environments with illumination color changes. In some computer vision algorithms, illumination changes are compensated by brightness correction. However, the compensation is insufficient for tracking in recent indoor environments with some types of fluorescent lights and LEDs. The color changes by illumination should be considered in color object tracking. This paper presents an object tracking method based on the gray world assumption (GWA) that relates to the color constancy in human vision. The average color in the vision is regarded as neutral gray in the GWA. In this method, the object color is defined as relative values from the average color in the frame of the image sequence, based on the GWA. Particle filter is used for robust tracking against sudden situation changes. We applied the proposed method to person tracking. Several experiments were examined for evaluation of the proposed method. The experimental results showed good feasibility of the proposed method.