Robust Color Tracking to Ambient Light Changes

Abstract

The use of color is a good technique for tracking objects. This technique has offered many advantages related to the robustness over occlusions, rotations, different geometries and scales. However, there are many studies that claimed that this technique generates excessive computational cost, and loss of reference due to the similar background color or loss of reference due to drastic changes in lighting. Therefore, there is a need to create a robust algorithm to prevent all the disadvantages mentioned above. This algorithm was mainly built on the analysis of the hue and saturation channels in the histogram based on the color space (hue, saturation and brightness - HSV). To resolve the disadvantages, an automatic recalculation of the histogram at each n frames has been developed. The algorithm has been enhanced to make it able to do all the process in real-time in large images up 1920x1080 pixels. The possible use of the algorithm is for controlling an automatic unmanned vehicle that will track a person.