Research On Security Monitoring Technology Of Intelligent Personnel Based On Image

With the rapid development of science and technology,people have higher and higher requirements for safety,especially in some practical industrial production scenario.But there are some specified safety codes,to ensure the safety of property loss and personal lives are inviolable which would be achieved through the video monitoring system to the appropriate regional or scenarios for all-weather monitoring.However,only relying on the traditional video monitoring system is inefficient.This method not only has high cost,but also needs to employ designated personnel to continuously watch the monitoring screen for supervision.With the increase of working time,the human perception of the environment will also decline,which is very prone to misjudgment and omission of illegal positioning.Therefore,it is very meaningful to design a set of intelligent personnel security monitoring system to meet the higher degree of automation.In this paper,common technologies in intelligent personnel security monitoring system are studied,including face detection algorithm,face recognition algorithm,recognition of staff safety clothes and safety helmet,and limited area intrusion detection.Based on the detection algorithm of the above parts,the intelligent operation of the whole video monitoring system is realized.Firstly,this paper studies the face recognition and personnel identification algorithm at the entrance of the security video surveillance system.Based on the target detection algorithm,Pyramid Box,and deep learning method,the accurate face detection is realized in complex situations such as small target,fuzzy and semi occlusion.Then the algorithm network is trained with a large open source face detection data set on the basis of pre training weight model.Dlib tool library and Euclidean distance threshold method are used to detect and classify the key feature points of human face,and the feasibility of the face recognition algorithm is verified by the self-made face database.Secondly,the network structure of pyramidbox detection algorithm with good detection performance is improved to realize the multi part detection of the head and body in the monitoring field of vision.Based on this,the detection algorithm of staff safety helmet is studied and verified.After the head and body regions are detected accurately,the two parts of the image are segmented and the color features are extracted respectively.The support vector machine is used to classify the feature information.By making a large number of data sets,the accuracy of detection and recognition is improved.Then the limited area intrusion detection algorithm is studied.Based on the good target detection accuracy of the improved pyramidbox algorithm,the intrusion detection function of the video surveillance system is realized through the location relationship between the target detection frame and the artificially delimited boundary of the limited area.Finally,the functions of detection and recognition of each part are integrated,and the human-computer interactive visual interface is developed to call each function and display the detection results.The feasibility and accuracy of the detection algorithm of each part are verified through experiments,and the expected goal is achieved.A set of security monitoring system with complete function and high intelligence is completed.