The Realization Of Human Gait Recognition By Using Acceleration Sensor

The realization of human identity recognition by using physiological or behavioralcharacteristicsis a very popular biometric identification technology now. Gait recognitionin biological characteristics identification is an emerging field, it uses the accelerationsignal from the wrist when people walking. The advantage of the gait recognition is thehuman gait feature can be stored in the electronic equipment in continuous identityverification and does not require the user’s always fit to identity verification. It hasimproved the identity authentication satisfaction and more effective guarantee informationsecurity. In order to improve the wearer comfort, also in order to reduce the productioncost of the system anddata operation cost, thispaper chooses threeaxis acceleration sensortorealizehumangait recognition.To achieve the body’s normal gait identification by using three-axis accelerationsensor, this paper used a method of extracting gait acceleration feature point based onacceleration sensor to realize different human gait certification. First of all, a hardwaremodel of human gait accelerationacquisition is built, which iscomposed of the MMA7455acceleration sensor and STC89C52microcontroller. Then selects10healthy users in theirchoice speed for three direction acceleration data acquisition by200Hz sampling rate.Through the serial port assistant ECOM, data is stored in the PC. The collected differentindividualacceleration data samples need be analyzed and processed. During data analysis,first calculates the relationship of test user’s pace, stride frequency and stride length; thenanalyses periodic acceleration signal. On the stage of data processing, first uses the FIRdigitalfilter and db5wavelet3layer decomposition to compare the signal denoising; thenuses the signal of the local maxima and local minima to finish cycle division, andstructures sample dimension combined dynamic structured network; finally, usingGaussian first order derivative function extracts each gait cycle feature points through thewavelet transform. During the recognition phase, first builds four Human Gaitdiscrimination system based on the feature points; then determines the sample featureinformation distortion distance based on dynamic time warping algorithm (DTW) to discriminate each level of human gait characteristics; finally,determines that whether thetest sample and the selected sample template match. According to the above method, thispaper basicallyachieved thepurpose of identification.