Research And Implementation Of Gait Recognition Of Multi-feature Fusion Based On Kinect

In recent years,in a complex and ever-changing international environment,terrorist attacks have occurred from time to time and people's demand for security has increased.In particular,how to accurately identify individual identities in crowded public places is an urgent problem to be solved.With the rapid development of information intelligence technology,gait recognition,as a new type of biometric technology,makes use of the inherent physiological characteristics or behavioral characteristics of human beings to identify individual identities based on the person's walking posture.It has the advantages of long distance,non-invasion,camouflage difficulties and low resolution.After a lot of research on gait recognition using Kinect skeleton data,this paper proposes a gait recognition method based on the multi-feature fusion of Kinect.The main steps of gait recognition are data preprocessing,extraction of gait features,and recognition algorithms.The main work and results are as follows:(1)Gait skeleton data preprocessing stage:The gait skeleton data preprocessing mainly solves through two aspects: First,the gait skeleton data visually checks whether the walking skeleton of the human has obvious deviation.The second is to detect whether the cycle of walking is unusual through the step length.(2)Gait feature extraction stage:After removing the noise from the bone data,in order to fully reflect the uniqueness of the individual,gait features are extracted from multiple angles,including both static and dynamic features.Using the intrinsic physiological structure of the human body,a rod-shaped skeleton model is proposed.According to a representative key gait,human pose features are extracted.Use the skeletal parameters to find the center of mass,and then calculate the distance from the center of mass to the main joint point to measure the change in the center of mass when walking,and the characteristics of hip joints reflecting the changes of the lower extremity joints while walking.(3)Gait recognition and fusion stage:After extracting gait features,the K-nearest neighbor algorithm based on dynamic time warping is used to match each gait feature.The matching scores that have their own characteristics are combined at the decision layer using the weighted addition rules to obtain the final decision value to identify the individual identity.A 90% recognition rate was achieved on the UPCV data set through simulation experiments.Verify that the method proposed in this paper has certain feasibility.Then implement a Kinect-based gait recognition system.The system takes as input the gait bone dataset collected by Kinect.The above-mentioned method is used to analyze and process the bone data and finally identify the individual's identity.The simulation experiment and the system implementation are to verify the feasibility and practicability of the gait recognition method based on the Kinect multi-feature fusion proposed in this paper,and to provide the basic role for the subsequent research and application.