Illumination invariant color model selection based on genetic algorithm in robot soccer

In this paper, we present a new approach to color recognition which is used in semi-autonomous robot soccer. Firstly, the new illumination invariant color space is defined based on the dichromatic reflection model. Secondly, we determine the most discriminating color model among a multidimensional set of color models by means of the genetic algorithm. Finally, we make use of the Gaussian model method for image segmentation. We successfully apply our new approach to semi-autonomous robot soccer game for the robot recognition. Experimental results demonstrate that the proposed approach achieves the high object recognition accuracy.