Research And Development Of Security System Based On Caps-Net Face Recognition

In public places with high security requirements such as airports,stations,and important square business districts.Effective security precautions,control,and monitoring of specific personnel are important means and guarantees for urban public safety and case detection.Security systems that rely on access control,manual inspections,intercom systems,etc.,are not only inefficient,but also prone to omission errors and increase hidden safety hazards due to fatigue of inspectors.It is more likely to cause queue congestion during holidays and peak periods,and it is easy to cause public dissatisfaction in terms of convenience and humanized experience.Therefore,automatic and efficient completion of face recognition,identity verification and security inspection tasks has become a strong demand for various security systems and face recognition systems.With the great progress and development of computer vision technology,the accomplishment of many tasks based on computer vision technology is higher than manual methods,which makes it very important to apply this technology to actual work and life.The main research contents of this article are as follows:1.Study the problems and specific needs of security project work;Model and analyze the system;Design the database;Develop the functions of each module;Provide identity authentication function,security service function,image service function,service management function,etc..2.Based on the microservice development framework,using HBase to support object storage and HBase’s good scalability,the security system is designed and developed.The system framework adopts mainstream development frameworks such as Spring,Spring Boot and Spring Cloud,to provide guarantee for system development efficiency,deployment operation and operation efficiency.The database adopts My SQL and HBase storage methods in order to achieve the requirements of fast response and operation search under big data.3.In view of the defects of the model of convolutional neural network series(CNNs),which is insensitive to image rotation and blur due to the loss of information in the pooling layer,this paper studies a slice-based recurrent neural network and capsule network(Capsule Network,Caps-Net)image feature extraction model,using the advantages of Caps-Net to retain information such as position and angle in the image,solve the internal knowledge representation and hierarchical modeling relationship in the neural unit,overcome the need of CNNs Learning different classification information for each different observation angle,which requires a large amount of data to generalize the problem.After an actual test by an institution,the security system has the characteristics of availability and reliability,which meets the security needs and project needs of the institution.