The Study Of Face Recognition Technology And Its Application In Safety Monitoring Of Oil And Gas Pumping Stations

With the rapid development of social economy,as the main energy artery of national economy,oil and gas pumping stations have made outstanding contributions to promoting regional economic development.However,in reality,the remote location of oil and gas pumping stations,the simple safety measures around them and the inability of personnel to monitor them all day,and the dark light inside the pumping stations and the prohibition of strong light lead to the blurred shooting effect of monitoring equipment and the gray face image,which seriously threaten the safety of oil and gas pumping stations.Therefore,the safety monitoring of oil and gas pumping stations has become a serious challenge to the national energy security supply.The application of face recognition technology in safety monitoring of oil and gas pumping stations can make full use of the existing face database resources and check the identity of personnel more conveniently,accurately and directly,which plays an important role in the construction and maintenance of oil and gas pumping stations.The main research work of this paper is as follows:(1)This paper Aiming at the problem that the face recognition rate decreases due to dim light in oil and gas pumping stations,a method of light processing based on decision tree and image enhancement is proposed.The results of experiments:Compared with traditional methods of light processing,such as linear transformation,logarithmic transformation,homomorphic filtering and quotient image,this method can significantly enhance each area of face image.The feature points of the domain play a very good role in the face recognition effect of the system.(2)The phenomenon of over-fitting of convolutional neural network in training samples is studied.The performance of CNN on training samples data sets is improved by adjusting the Super-parameters constantly through experiments,and the appropriate Dropout value is found to solve the problem of over-fitting.Then the convolutional neural network is established to validate and analyze the face recognition of the face images processed by decision tree and image enhancement illumination algorithm.Compared with the traditional illumination processing methods,such as linear transformation,Gamma correction and histogram equalization,the proposed decision tree and image enhancement method can improve the recognition rate by 3.01%,2.66%and 1.38%respectively.At the same time,the face recognition method based on convolution neural network in the system has 91.11%compared with the traditional face recognition method and more suitable for face recognition.(3)To meet the needs of oil and gas pumping stations,a prototype of face recognition system for safety monitoring of oil and gas pumping stations is designed and developed.Compared with other systems,the system has the advantages of good portability,low cost,good night monitoring effect and long service life.the prototype of the system mainly includes image comparison,video recognition and face database management functions.Experiments show that the prototype of face recognition system of oil and gas pumping station has good face recognition effect and plays a good role in safety monitoring.