Research On Static Gesture Recognition Based On Kinect Sensor

With the rapid development of science and computer technology,hardware equipment 's constantly updated,the interactive application between mechanical equipment and human is exponential growing,human-computer interaction technology has begun to become today's technology research hotspot.With the deepening of the research,the dominant has been changed from the computer to the user.At present,scholars are working on a user-centered,natural and harmonious human-computer interaction technology.Gestures are natural,direct,and user-friendly ways of interacting,gesture interaction has the advantages of fast,non-contact and easy to understand,therefore,the research on gesture recognition can enhance the user's immersion experience and make the interactive experience more intuitive and comfortable.Gesture recognition plays an very important role in sign language recognition,human-computer interaction,robot control,intelligent monitoring,visual environment operation and other fields,the research on gesture recognition is an indispensable part of the human-computer natural communication.At present,researches on gesture recognition at home and abroad are mainly divided into three categories: two-dimensional hand recognition,two-dimensional gesture recognition and three-dimensional gesture recognition,the first and the second are entirely two-dimensional,and the third contains depth information.At first,the researchers mainly used the direct detection of machine equipment to obtain the spatial information of human hands and joints,so as to identify the meaning of gestures.In 1983,the introduction of data gloves greatly improved the accuracy and stability of gesture recognition,but there were some disadvantages,such as limited naturalness of gestures and few identifiable gestures.In recent years,the technology of depth camera is gradually emerging.With the assistance of depth data,new methods of gesture recognition based on vision information is in an endless stream,which effectively promotes the practical process of gesture recognition system based on vision.but there are still some problems such as low recognition accuracy in complex background and insufficient generalization ability of the model.Kinect can capture three major data information: deep data stream,color data stream and original audio data.Kinect has many advantages such as powerful function,low cost,convenient development,etc.In view of this,this paper proposes a method to identify static gestures based on Kinect data,and mainly studies the followingaspects:(1)Design ten Static gestures based on Kinect data.In this study,ten gestures representing 0~9 were designed for experiments.Gesture are natural,common,and strive to make the experimental results more reliable and comprehensive,contrast obviously and analysis accurately.(2)Combined with color,depth data and skeleton information of Kinect,an algorithm of static gesture recognition was proposed.In this paper,the current research status of gesture recognition at home and abroad is deeply understood and studied,and relevant algorithms are compared experimentally.Aiming at the characteristics of variable gesture recognition,complex background and high real-time requirement,the advantages and disadvantages of existing algorithms are summarized,and a set of static gesture recognition algorithm based on Kinect data is proposed.The recognition algorithm include seven parts: data acquisition,preprocessing,hand location,background separation,hand extraction and gesture recognition.The static gesture recognition based on Kinect data is proposed that can ensure the correctness and real-time performance of the recognition.(3)Experimental verification.The static gesture recognition software based on Kinect data was designed and developed.the gesture recognition interface was built to realize the functions of data input,gesture recognition and result output.The efficiency and accuracy of the algorithm are proved by experimental comparison and result analysis.It provides a reference for future research on static gesture recognition.