Application Of Human Identity Recognition And Posture Recognition Based On Kinect Using In Smart Home

With the development of information technology,the requirement of effective authentication and human identity recognition and posture recognition becomes higher and higher in time attendance system and smart home.The traditional attendance system used manual registration or swiping cards,which may result in misappropriation and loss.It could not ensure uniqueness and error-free.Considering that most smart home products now on the market only stayed on a slow level of multi-channel control of home appliances and had a single product function.Therefore,this paper introduced Kinect sensor and proposed a human identity recognition and posture recognition system based on the human skeleton structure.The system used STM32 as the control core.Combined with SVM classifier and the Internet of Things technology.The system formed a more perfect identity recognition based on traditional identity recognition system.It also could be used in enterprise time attendance system as a second choice.Meanwhile,the posture recognition was applied in intelligent control of home appliances,and the user could complete the control of home appliances without a third-party medium such as a mobile phone.The system was designed with human face recognition and opening the door based on the structure of bones,strangers alarms,smart window and other security facilities to ensure family safety.After Experimental Verification,The security capability and intellective level of today smart home was raised by employing this system.This article focused on the composition of the system,the application of Kinect somatosensory technology,and the realization of a human identification system and a gesture recognition system based on human skeleton data.The system first used Kinect programming to extract human skeleton information.This method used the distance between each two skeletal joints as the feature.The improved SVM classifier was used to filter the features,and finally extracted 16 optimal features for recognition judgment.Then combined with the human skeleton features,Kinect depth image and color image to completed posture recognition.Matching the sensors of different home appliances with different character poses,and finally achieved the control of home appliances through gesture recognition.Specific study of the human skeleton recognition from the following modules:(1)Human identification system platform construction.This system used Kinect3D camera to perform tracking and recognition of human skeletons.The Kinect real-time character recognition system was developed based on Visual Studio 2013 WPF platform using C#language.The system included bone data entry,training model,adjustment,parameters,person recognition,posture recognition and real-time character image display.(2)Character identification based on Kinect.Using Kinect to obtain skeleton information,the distance between 31 effective joints were extracted as the features.SVM was used to classify the features training,and 15 optimal features were selected for indentity recognition;(3)Character posture recognition based on Kinect.Using Kinect Studio and Visual Gesture Builder for samples collecting and training,eventually achieved the control of home appliances through gesture recognition.(4)Design and implementation of smart home system.Including Internet of Things,client,server setup,online real-time character recognition system.Through Matlab simulation and testing on real users,the results show的that the character recognition and gesture recognition based on human skeletal data system had a good accuracy and could be used in intelligent home products.