Research On Dynamic Gesture Recognition Based On Data Glove

As one of the hotspots in the research field of human-computer interaction technology,gesture recognition has been deeply studied by domestic and foreign research scholars,and it has also been widely used in virtual reality and artificial intelligence.Gesture recognition is divided into static gesture recognition and dynamic gesture recognition according to whether the hand produces motion;according to the different ways of acquiring hand gestures,it is divided into gesture recognition based on non-contact sensor and gesture recognition based on contact sensor.With the popularization of sensors,gesture recognition based on contact sensors has developed rapidly,especially gesture recognition based on data gloves.Data gloves can more intuitively obtain three-dimensional spatial information of hand gestures by using multiple sensors.Constrained by the surrounding environment,compared with gesture recognition based on contactless sensors,the research and application value of gesture recognition based on 'data gloves is higher.At present,the research work on gesture recognition mainly focuses on the research of recognition algorithms,ignoring the user-centric concept in human-computer interaction,which limits the practical application of recognition algorithms in gesture recognition.This article summarizes the current knowledge of gesture recognition algorithms,combined with the application of gesture recognition technology in virtual reality training and assembly scenes,designed a set of user-centric gesture recognition methods.First,in order to ensure that users have a more natural and immersive experience in virtual reality scenes,this article refers to the Chinese sign language manual,the distribution of human hand bone joints and data glove sensors to predefine gestures;The advantages and disadvantages of the data,the gesture sample training set is optimized,and the k-means algorithm is sensitive to k value and centroid selection and the timing problem of dynamic gestures,a time-domain double k-means gesture is proposed Feature extraction method;for intercepting effective gestures,start and stop labels are set for each gesture;Several classic recognition algorithms are analyzed,a DTW method based on feature templates and with thresholds is proposed,and the recognition method is verified in virtual reality training and assembly scenarios.