Research On Gesture Recognition Based On Kinect And Its Application In Multimedia Assisted Teaching

In21century, education is devoted to intelligent developnent. Intelligent classroomis one of the most representative achievements of intelligent education.It's amodernization classroom, centered on human-computer interaction and relying onintelligence technology. It realizes interaction via gesture recognition. Usually there aremany approaches to gesture recognition and one of them is depending on machine visionwhich is adopted in this paper. Kinect camera is a somatosensory device developed byMicrosoft,which can well acquire action of human body and voice to realizehuman-computer interaction. This paper introduces a scheme based on Kinect and visualprocessing technology to achieve recognition of static and dynamic gestures inexperiment, which finally attains the purpose of assisting teaching.The main research of this paper is shown as follows: Firstly the method ofmeasurement gets improved and the internal parameters and distortion parameters incamera are attained,which is necessary for gesture recognition. Secondly acquired depthinformation is processed with the measurements of linear transformation and contourextraction.Further gesture region is extracted by threshold value method in combinationwith principle of nearest neighbor,distance metric function of the edge invariant momentof hand area is adopted to obtain static gesture recognition.Next the potential dynamicdetection and anti-shake is used to deal with three-dimensional coordinates of hand andthen chain code discretization is employed to extract dynamic gestures. The commonused algorithm of HMM in speech recognition is adopted to recognize dynamicgestures.Finally gesture recognition system is constructed based on gesture recognitionand a scheme is designed to apply the constructed system to multimedia education.Further the designed scheme gets implemented in real-time teaching.The proposed scheme can effectively identify static and dynamic gestures. Theexperimental results show static gesture recognition can perform well under theconditions of complex scene and changeable illumination and the undefined dynamicgestures are correctly rejected,which proves the whole system performs robustly.However, there is some further work to improve the deficiencies of this scheme, whichcan finally enhance the system usability and reduce error rate of gesture recognition.