Research On The Detection And Recognition Of Motion Track

In recent years,with the rapid development of Micro-Electro-Mechanical Systems technology and wearable technology,more and more intelligent wearable products aimed at detecting the movement of objects emerge in people's lives.It becomes feasible for data acquisition and recording the movement of objects on wearable devices,such as linear acceleration,rotational angular velocity,etc.However,the motion information data is too large during a continuous long range,and the details can not describe the motion of the image,which is not conducive to the user analysis,observation and extraction of valuable information.Therefore,this paper aims to realize a trajectory and identify objects specific category method used to effectively detect.The method can not only extract the moving attitude and trajectory of the object,but also can recognize different types of movement.Besides,it can be derived from a variety of different applications by using this technique in conjunction with specific sports,and the paper implemented the motion badminton category so as to complete their specific application layout.In the sports field for badminton,collection by multi-sensor acceleration,angular velocity data badminton and other sports,it can be achieved on the racket attitude and trajectory solver extraction,which can restore the entire swing process batting-based data acquisition,and be able to identify ways in which the ball belongs to the stroke trajectory.The final details of badminton shots through 3D animation and motion parameters calculated feedback to the user,the user can response results for targeted training in order to improve the technological level of the sport.Firstly,acceleration,gyroscope,geomagnetic data were conducted posture resolving,increasing the system based on the traditional Kalman filter quaternion solution method for state variables,besides,improving the method of calculation Solution effect by introducing a gyroscope errors to.By contrasting the experiment data,we can know that posture angle output for multi-state quantity solvers posture are more stable than traditional posture resolving.Extraction trajectory based on the posture resolving,for purification numerical integration of the acceleration signal,by fitting a polynomial to eliminate cumulative errors produced during integration,which makes extraction more accurate trajectory.Secondly,according to the track extraction results based on each set of motion data posture,system extracts the maximum amplitude and the attitude angle change speed as the SVM feature vectors,by constructing multi-classification model to achieve the trajectory recognition.In this paper,badminton racquet track statistics and analysis choose the appropriate system parameters in the process of building multi-classification model,do it is possible to accurately identify eight different swing,the experiments validate the effect of the method.Finally,trajectory and attitude of badminton in the form of 3D animation presented to the user by using visualization software swing,at the meanwhile,it is able to swing the ball way to accurately identify trace.In addition,the feedback of relevant technical parameters will also be obtained,so the user can improve the technological level of the sport according to the feedback results by targeted training.