Continuous Gestures Recognition Without Contact Based On Wi-Fi Signal And Sequence Alignment

People use gestures to express meaning in daily life,and more need to use continuous gestures.At present,in the research of wireless sensing,continuous gesture recognition is still a difficult problem.This paper proposes a WScg algorithm based on wireless signal technology and biological sequence alignment technology for continuous gesture recognition.The main works of the thesis are listed as follows:In this paper,the characteristics of continuous gesture and single gesture are studied firstly.And then,referring to the processing method of decomposing sound signal into syllables for recognizing,The processing time of Channel State Information(CSI)is set to 0.1 seconds to determine the gesture to be recognized in the actual continuous gesture.It provides accurate start time for continuous gesture recognition algorithm.Furthermore,this paper designs a dynamic and static separation method of CSI gesture data sequence to solve the problem that training data standards cannot be unified in training samples.In this method,support vector machine(SVM)is trained by using the data separated from experience knowledge,and then a unified standard training data is produced by separating all the training data from the static and dynamic ones.Experiments show that the dynamic and static separation method can improve the recognition accuracy of the algorithm by 20%.On the basis of the above research,based on Smith-Waterman algorithm in biological information sequence alignment,this paper designs WScg sequence alignment method to achieve accurate gesture recognition.WScg sequence alignment method is used to compare and score the information of each moment synthesizing its front and back moments,to determine which kind of gesture start time is at that moment,and to get gesture categories and gesture start time.Experiments show that the WScg algorithm can achieve continuous gesture recognition,with an average recognition accuracy of 87.8%.