Multi-feature criteria with fuzzy logic pattern recognition for hand gesture classification

This paper presents a method of hand gesture sequence recognition through a multi-feature criteria based on a predefined area in three dimensional space, the presence of skin color, and non-motion for a defined period of time using a depth camera and depth information. Hand gesture recognition is an active topic of research and development for human computer interfaces. Hand gestures naturally consist of movement and paused motion. This technique focuses on the non-motion features of a hand gesture. It is easy for a human being to recognize the meaning of a common hand gesture but it is difficult for a computer to accomplish this same task. A systematic method was developed to distinguish paused motion from hand movements so that pattern recognition techniques can be effectively utilized to interpret the gesture. The approach was successfully validated by experiments.