Research On Gesture Recognition Model And Its Application Based On Wear Sensing Perception

With the development of virtual reality technology,smart home technology and immersive learning technology,the existing interactive control methods cannot meet the needs of increasingly rich interactive scenes.Gesture recognition technology could make full use of the characteristics of flexible hands and release the interactive nature of hands,which is an important supplement to the interactive mode.However,the current research on gesture recognition technology mainly focuses on video gesture recognition.In contrast,there are a few studies on sensor-based gesture recognition.Moreover,most of the researches need to use professional equipment or self-made equipment,which is not easy to use and promote.This paper aims to provide an efficient and accurate gesture recognition and real-time interaction system based on wear sensing perception by utilizing commercial smart wearable devices,gesture interaction theory and machine learning method.In this study,commercial smart bracelet is used as experimental equipment to collect accelerometer data and gyroscope data of hand raising gestures in four common scenarios.The classification model is trained through data preprocessing,data segmentation,feature extraction and classification model training.Among them,the improved real-time gesture segmentation algorithm that utilizes sliding window with variable step size based on the feedback of classification results combined with the double thresholds can give consideration to real-time performance of interaction and the accuracy of recognition.In this paper,Flappy bird game and smart light bulb are introduced to evaluate the real-time interactive effect.The bird in the game is controlled to fly upward and the color of the light bulb is controlled to change by raising the hand gesture.The system could realize high-accuracy static recognition and smooth real-time interactive control for simple gestures,which not only enriches the human-computer interaction mode,but also fully exploits the data generated by the wearable device,bringing new breakthroughs for the wearable industry development.At the same time,it will provide new solutions for smart home control,VR interaction,game control,Parkinson patient hand function rehabilitation,deaf-mute sign language recognition,gesture input and other application scenarios.