Abstract
The tracking by correlation is the method which has been chosen by the creators of the magic-board for its performances. They have adapted this method to their needs and realized a finger tracking in real time. In this paper, we propose the adoption of a simple and efficient tracking method applied to the magic-board. This method was inspired from the algorithm of Frederic Jurie & Michel Dhome. Knowing the position of the target in the first image, we can estimate the position of this target in the subsequent images by comparing the grey level values of the target pattern with the grey level values of the predicted regions (search region). The relation between the intensity variations and the position variations is calculated during the offline training stage. The proposed algorithm allows a better flexibility to the user of the magic-board. Our system can track the user's finger at any position in real time.