Visual Feature Extraction And Establishment Of Visual Tags In The Intelligent Visual Internet Of Things

Intelligent visual Internet of things(IVIOT)is a kind of Internet of things with the function of visual perception,which is consist by four parts:intelligent vision sensors,intelligent visual information transmission,intelligent visual information processing and internet of things application.Intelligent visual networking is the escalation of the Internet of things,it uses image processing technology and computer vision technology to identify,position,track the content intelligently,which is transmitted by image sensor.It can help us to manage people,vehicles and objects intelligently.One of most important technologies of intelligent visual Internet of things is the intelligent visual tag system.It can identify the contents of the video or image,understanding and classification,and mark it,then display the label's information of the object which is already identified.In this paper,a visual label system based on human,vehicle and object is designed and implemented according to the needs of the subject.The system mainly includes the human recognition module,the vehicle identification module and the object recognition module.In this system,users could select a picture according to their needs,then the system will automatically identify the contents of the picture,and show other pictures related to it and its information tags.In the human recognition module,this paper adopts the face recognition technology based on principal component analysis(PCA)algorithm and support vector machine(SVM)algorithm for face recognition.Firstly,the PCA algorithm is used for feature extraction and dimension reduction,then the SVM algorithm is used to classify and recognize.In the vehicle identification module,we identify the vehicle license plate used color-based license plate recognition method.Realizing license plate positioning,license plate correction,character segmentation,character recognition step by step for the vehicle in the image.Finally the license plate number is identified.In the object recognition module,this paper adopts the object recognition method based on convolution neural network.This article will identify the same kind of objects(this paper takes cups as an example).Finally,the three modules are combined into one system,which makes people,vehicles and objects correspond to each other.The experimental results show that the system has good performance and can meet the basic needs of users.