Dynamic contour matching for hand gesture recognition from monocular image

Hand gestures are used widely in communication. An important example is using in the sign languages. Many hand gesture silhouettes are the part of other hand gesture silhouettes. For example, V sign gesture is a part of the high five gesture, because we can create high five gesture silhouettes from the V sign gesture silhouettes by extending the other three fingers. Here we propose the partial contour matching algorithm for gesture classification. Our classification is to find the deepest gesture in a tree such that none of more its children are the part of a sample silhouette.