Gait Detection System Based On Visual Sensing

Gait analysis,started in the 90 s of last century,provides an objective and quantitative evaluation for human walking function.Based on parameters(time,geometry,mechanics and electromyography,etc.)of human walking.A kinematic observation and dynamic analysis was carried on limbs and joint activities of human walking.An abnormal gait analysis help doctors to pre – judgment the related disease and play an auxiliary medical assistance.At present,intelligent wearable gait detection,using the contact measurement method,dynamic and electromyographic parameters(time-space,gait cycle,lower limb joint angles,myoelectric activity,focus and energy change,etc.)can be obtained to complete gait analysis.A gait detection system based on visual sensor was designed in this paper.Using non – contact visual imaging technology to capture image of human gait,the depth human skeletal analysis was carried out to extract the characteristic parameters of gait.The fusion of DTW(Dynamic Time Warping)algorithm and gait parameters was used to achieve gait analysis.Firstly,the Kinect sensor was used to gain the dynamic skeleton parameter and the gait feature was extracted to establish the template of gait.Secondly,the DTW algorithm was used to match and identify the gait pattern.Then,with the acquisition of skeletal fulcrum data,using the approach of event to divide the human gait to calculate the gait parameters,and using the AHP and the reasonably weights to evaluate human gait to improve the accuracy of gait recognition.A test platform was set up for gait detection system,selecting 150 gait sets as test samples to study the human gait.The test results show that using the fusion of DTW and gait parameters to analysis gait,the human gait recognition rate can reach 86.6% is higher than using DTW algorithm,which gait recognition rate is about 79% under the same experimental conditions.The improved algorithm improves the recognition rate of human gait and validates the effectiveness of the algorithm.