3D Depth Information Based 2D Low-Complexity Hand Posture and Gesture Recognition Design for Human Computer Interactions

Abstract

Owing to the smart applications of human-computer interaction (HCI), the hand posture and gesture recognition technologies have acquired more and more attentions recently. In this work, the 2D low-complexity hand gesture identification technology is proposed based on 3D depth information. The proposed design overcomes the difficulties to separate the integrated palm region from the complex background. First, the proposed system uses the 3D depth camera to obtain the depth information. Then the system uses the depth image of the palm area to extract the contour and features of hand. By the contour and features of hand from geometric relationship, the proposed design recognizes several useful hand postures and gestures effectively. In our experiments, the recognition accuracy of hand gestures can be up to 98%. By implementing with the 4-core PC (Intel i7-4720, 2.6GHz, 8GB RAM), the average processing speed is up to 15 frames per second.