Fast fingertips positioning based on distance-based feature pixels

This paper presents a robust new method to fast and precisely position the fingertip of all open fingers on a skin color detection image. Based on a set of distance-based feature pixels, which are extracted on both the connected component labeling image and distance transform image, a new principle to select the hand region and a new principle to determine the center of the hand in an image have been proposed in this paper. With a clustering procedure and three refining processes on the distance-based feature pixels, the final set of the distance-based feature pixels almost includes only the feature pixels on the open fingers. As the result, the fingertip positions can be determined on this set more easily through a calculation of the Euclidean distances from the distance-based feature pixels to the center of the hand. This proposed method can be carried out fast and accurately, even when the skin color detection image includes a hand, faces and some noise.