Research On Gait Recognition And Gait Symmetry Based On Deterministic Learning Theory

Gait is a gesture that people show when walking,is a complex movement process.Gait recognition is an emerging biometrics technology.By objectively recording and assessing the human body walking style effectively,biometrics and kinematics are used to identify or physiologically,pathologically and psychologically Feature detection.Compared with the traditional biometrics,gait recognition has the advantages of non-intrusive,difficult to hide and imitate.With the continuous research in the field of computer vision,the achievements have been widely used in biometrics,security monitoring,rehabilitation robots,etc.field.In this paper,Kinect2.0 sensor is used as an experimental device,focusing on the design and implementation of gait recognition system based on deterministic learning theory,and the related content of gait symmetry.Deterministic learning theory is a new theory of machine learning that uses Radial Basis Function(RBF)neural network to acquire,express,store and re-use knowledge in an unknown dynamic environment.It can locally and accurately recognize the periodic or regression system,and save the dynamic knowledge learned in the process in the form of the weight of RBF neural network.Then,a set of dynamic estimators is constructed by using the RBF neural network to represent the training pattern,a set of errors is generated as a measure of the similarity between the test pattern and the training pattern,and the test pattern is quickly identified according to the principle of minimum error.Based on the deterministic learning theory,based on the work of predecessors,this paper makes an exploration and research on gait recognition system in practical application.In this paper,a real-time gait recognition system is designed and developed.Kinect sensor is used as a new method for gait feature extraction.Based on this,a real-time gait recognition system is designed based on MATLAB GUI and C # GUI.The main functions of the system include gait skeleton Flow data acquisition,data filtering,gait learning and training,gait recognition,gait three-dimensional dynamic chart display and residual analysis,and other functions to achieve automatic gait feature recognition and online recognition of results obtained through self-built database validation The accuracy and rapidity of gait recognition proves the systematic application value of the system.Based on the deterministic learning theory,the gait symmetry is quantified.Based on the residuals in the gait recognition system,the dynamic and static information contained in the experimental data is preserved,and the slow A new gait symmetry measurement method was proposed based on the comparison of the gait symmetry between the left and right lower extremities in walking,walking at high speed,walking at high speed,walking under weight,walking at different shoes on the left and right feet,and deliberate asymmetrical walking.