Posture independent model for hand detection and tracking

Human Computer Interaction takes the attention of many researches all over the world, researches try to invent more natural and intuitive interaction ways between humans and computers. However, people depend mainly on their hands to interact with surrounding objects in everyday life. In this paper we present a novel appearance based method to recognize and track human hand in an unknown environment without any constraints on hand posture and with no need to wear any additional devices. Our method can distinguish between hand and other moving objects, by using proposed posture independent hand features. Results show that our method can recognize and track human hand at real-time with 96% accuracy rate in normal indoors light.