Human-computer Interaction System Based on Visual Gesture Recognition

Human-computer interaction system is the medium for communicating and transmitting information between people and computers. With the rapid development of computer technology, traditional human computer interaction technologies such as mouse and keyboard, have not met the needs of the development of the times. Instead, people need another human-computer interaction technology which is faster, more natural and comfortable. Gesture-based human-computer interaction is one of the most important technologies in human-computer interaction system. There are problems remained in traditional gesture recognition methods, such as low recognition accuracy and complicated recognition process. In view of the defects above, this paper proposes a gesture recognition algorithm based on deep learning. The algorithm detects joint features of the gesture quickly through gesture estimation and classifies joint feature maps by using convolution neural network, which overcomes the difficulties of segmenting gesture images in complex background and improves the accuracy of recognition results. The experimental results indicate that the method has high recognition accuracy for various gestures at different scales, which reaches 98%. Finally, a human-computer interaction system is designed based on the algorithm, and the application of gesture recognition in the human-computer interaction system is demonstrated.