Research On Unattended Well Site Security Video Surveillance System Based On Computer Vision

With the vigorous development of oilfield information construction.China has established a large number of unattended wells.Most unattended wells are located in remote areas,and the traditional security video surveillance system cannot guarantee its production and property safety.So it is very important to speed up the information construction and intelligentization reconstruction of the security video surveillance system.Based on the existing research results of video surveillance system and computer vision technology,combined with the environmental characteristics of unattended wells,this paper proposes a security video monitoring system which can automatically monitor the intrusion behavior of unattended wells,completed security video surveillance system of the overall scheme design,including the selection of the cameras,the spot POE power supply,management and storage the scene video information,the remote video transmission mode and remote video monitoring client software with motion target detection and pedestrian detection.Based on the environmental characteristics of unattended wells,the corresponding simulation experiment platform was built.In the aspect of moving target detection,this paper analyzes the characteristics of the frame difference method and the background difference method in extracting the moving target under the condition of fixed camera,and puts forward the improved background difference method in combination with the advantages of the two.Under the condition of global illumination change.The comparison results show that the moving object detection effect based on the improved background difference method is better than the background difference method based on the mixed Gaussian model.In the aspect of pedestrian detection,using the principal component analysis method to reduce the HOG pedestrian description features,and with the support of OpenCv,based on the INRIA pedestrian database and self-collected well field realism as the sample set,the pedestrian detection classifier training is carried out by support vector machine,the experiment shows that the classifier has high detection rate.Finally,EmguCv is used to encapsulate the improved background difference and pedestrian detection algorithm to PC-based NVR client software written in C#language,at the same time the software also has the function of video information management,video information storage,video information analysis,remote PTZ control,remote parameter configuration and so on.The relevant test result of the simulation experimental platform shows that,this unattended security video surveillance system has feasibility.It is of great significance to improve the safety of the traditional video surveillance by the operator,to prevent people deliberately invasion,and to ensure the production and property safety of unattended well site.