Research Of Gesture Interaction Based On Inertial Sensor

The technology of controlling the intelligent user interface has been paid more and more attention.In the field of human-computer interaction,gesture action can be used as a new model.With the characteristics of intuition,nature and variety,gesture interaction can give people a more intuitive and comfortable natural interaction experience.The traditional method of gesture interaction based on inertial sensing focuses on how to make the person-independent gesture recognition method more robust,and obtain faster dynamic response.But the traditional method does not take into account the regularity and validity of the sample set in the algorithm,and affect the recognition accuracy of the algorithm to a certain extent.At the same time,when the gesture set is more complicated and the types of gestures are increased,the traditional method is more susceptible to redundant information and noise information in the gesture signal,resulting in misjudgment of the gesture category.In view of the shortcomings and disadvantages of traditional methods,in order to reduce the complexity of the calculation and improve the accuracy of hand gesture,the improved gesture interaction method based on inertial sensing is proposed.Experimental results show that the computational time of this method is at least a 25% decrease comparing with the traditional DTW algorithm,and the overall average recognition accuracy is between 96.7% and 98.84%,which is obviously better than other traditional algorithms.This paper is mainly focused on the following three improved research aspects.1.Aiming at the construction of sample set in traditional methods,in order to improve the non-regularity of sample selection,a sample clustering training method based on CDTW algorithm is proposed in this paper.This method can not only overcome the speed difference between different individual gestures,but also improve the regularity and validity of the sample dictionary.The typical sample obtained by cluster clustering compresses the size of the sample set to a certain extent,and more importantly,contains the typical characteristics of different gestures from different individuals,so that the method has stronger individual adaptability.2.The problem of heavy computation in the traditional gesture interaction method process of recognizing can be improved by the method of clustering the sample.At the same time,this paper also proposes the principle of main axis classification.In the operation,the test gesture sequence matches with the same main axis samples,greatly reducing online template matching time complexity of the gesture recognition process.3.This paper proposes a method of using the compression sensing method as a recognition way in gesture interaction considering that the traditional method is more susceptible to the influence of redundant information and noise information in the gesture signal,which causes the misunderstanding of the gesture category.The use of this dimensionality reduction method not only reduces the amount of computation,but also can restore the gesture signal without distortion,retain important features,and then improve the recognition rate of the algorithm to get the recognition result.