Gesture Recognition Based On Probability Matrix Model Research

A gesture recognition technology–based image or vision has a high recognition rateand good advantages of real-time effects, it is widely used in intelligent systems and otherareas, but the surrounding environment light and background image intake process andeven shooting angle dependence restrict the development of space in the future. With therapid development of sensor technology, the acceleration sensor-based gesture recognitionis becoming a hot research in the field. Gesture recognition technology–basedacceleration sensor is still in its infancy and needs to be further enhanced in therecognition rate and real-time effects.On the basis of the research status at home and abroad, an entirely new gesturerecognition method based on probability matrix model has been designed in this paper.First of all,3D-gestures data signals are collected by acceleration sensor (AMI602). Andthen the acceleration signal is sent to a txt file saved to a PC through the serial port. Thehexadecimal data signal is calculated into decimal acceleration values.Secondly, against the recognition method designed in this paper and the accelerationsignal characteristics, the gestures acceleration data signal is pretreated, it mainly includesa filtering process to remove an interference signal, detecting the effective action of thesliding window of data detection, adjusting the sampling frequency of the interpolationprocess, and normalization of the size of the entire amplitude normalization process.Finally, this design mainly researched the algorithm of gesture recognition. Bylearning and summarizing those existing gesture recognition methods, the discrete HiddenMarkov model would be established to identify gestures recognition. The gesturerecognition method based on probability matrix model of interval distribution is proposedin this paper, and it is elaborated in five aspects of extracting from the feature sequence,model building, model initialization, model optimization and recognition process.The system applys the MATLAB environment for software platform, the feasibilityof the algorithm is verified by simulation. In experiments, the average recognition rate of 10custom single-stroke gestures is90.2%, and the Arab digital is84.4%.