Design And Implementation Of Non-Contact Intelligent Human-Computer Interaction System Based On Kinect

With the rapid development of artificial intelligence,the user’s demand for human-computer interaction continues to increase.Traditional keyboard,mouse and other interactive methods are becoming more and more difficult to meet the needs of natural interaction.People urgently need to develop an intelligent interactive system based on multi-biological recognition technology.This paper aims at the problem of single interaction mode and poor contact interaction experience in the actual application of multi-biological recognition interaction technology.Kinect is used as the human-computer interaction medium,and the system that uses face recognition as security authentication and gesture and voice as interactive control is proposed.The design and implementation of a non-contact intelligent human-computer interaction system based on Kinect.The research content of this paper is as follows:First of all,in view of the slow running speed and low face recognition rate of face authentication in the human-computer interaction system,a deep learning acceleration scheme Open VINO is proposed to optimize the deployment of face detection and feature extraction models,and to integrate template matching and support.Vector machine method for face recognition.Experimental results show that the algorithm not only reduces the image processing time,can run the model in real time,but also solves the interference of undefined faces,and improves the response time and recognition rate of face recognition.Secondly,in view of the poor recognition accuracy and real-time performance of gesture interaction control in the human-computer interaction system,it is proposed to study from two aspects of static gesture and dynamic gesture:(1)In the static gesture recognition,the use of directional gradient histogram features and Gray space histogram features represent gesture feature information,and static gesture recognition is completed by weighted fusion support vector machine and integrated distinguishing random fern algorithm;(2)In dynamic gesture recognition,it is proposed to construct by the angle and distance features of specific human joint points Gesture action feature expression,using Mahalanobis distance for dynamic gesture template matching.Experimental results show that the algorithm effectively improves the accuracy and real-time performance of static and dynamic gesture recognition,and then realizes rapid gesture recognition,and has stronger robustness.In addition,in response to the problem that the voice interaction in the Kinect human-computer interaction system does not support Chinese voice recognition,it is proposed to embed Microsoft Speech Platform Runtime v11 into the interactive system to achieve dual voice control in Chinese and English,and provide a wealth of human-computer interaction methods.Finally,a non-contact intelligent human-computer interaction system based on Kinect is designed and implemented.The interactive system performs modular channel design and collusion for face recognition,gesture recognition and voice recognition according to logical interaction requirements,and controls the video player through human-computer interaction.The experimental results show that the overall operation of the system is stable and smooth,the user experience is good,and it has practical promotion significance.