Study On Intelligent Monitoring System Of Field Oil Pipeline Based On Computer Vision

Due to its safety,economy and speed,the oil pipeline has been paid attention to and developed rapidly in China in recent years.It is of great significance to develop the oil pipeline transportation industry to allocate crude oil resources within the petrochemical industry,accelerate the utilization of domestic and international crude oil resources,reduce operating costs,and improve the operational quality of the domestic economy.In the long-distance transmission process of crude oil,there are problems of oil pipeline damage caused by road construction and hidden dangers of illegal molecules punching and stealing oil.Therefore,supervision of oil pipelines is indispensable.Aiming at the problems of traditional manual supervision,such as large workload,low efficiency and poor real-time performance,this thesis proposes a computer vision-based intelligent monitoring method for oil pipelines.The research content includes the following aspects:(1)In the aspect of image dehazing treatment,there is a large amount of fogging on the captured image,which causes the target recognition degree to decrease.The method of improving the identification accuracy by defogging algorithm is studied,and the dark channel based on mean filtering is proposed.Image defogging algorithm.The process of degrading the foggy image is analyzed,and the foggy image degradation model is established.The dark channel prior law is used to restore the fog-free image,which enhances the sharpness and contrast of the foggy image.Aiming at the shortcomings of the dark channel prior algorithm,the algorithm is improved.The mean filtering is used to estimate the ambient light and the global atmospheric light,which effectively solves the problem of the restored fog-free image darkness,color distortion and white edge.Through the contrast experiment,the improved algorithm can get better dehazing effect.(2)In the aspect of image background modeling,the image update time interval is long,which causes the background change to be large,which causes the low accuracy of the background image.The image background modeling and updating method is studied,and the statistical mean background model is proposed.Image background modeling algorithm.In order to improve the accuracy of the background image,the mean background model algorithm is improved.The background image is accurately constructed by statistical mean method,and the background image sequence is updated to construct a background image that adapts to the environment change in real time.Through comparison experiments,it is verified that the improved background modeling algorithm proposed in this paper can effectively improve the accuracy of background images and has certain advantages in adapting to background discontinuities.(3)In the aspect of target detection and anomaly classification,aiming at the problem of low accuracy of target detection caused by illumination changes,an adaptive threshold background subtraction algorithm is proposed to effectively eliminate the illumination change problem and achieve better target detection results.Aiming at the problem of detecting target shape distortion caused by target shadow,an improved algorithm based on HSV color space is proposed to solve the problem of incomplete shadow removal.Using the aspect ratio of the foreground target,combining the moving speed to determine the target type,and setting the alarm priority for different types of targets,the problem of target type classification and abnormal behavior judgment is better solved.In this thesis,the target image is detected by the image sequence collected by the actual oil pipeline intelligent monitoring system.The test results show that the system has high accuracy and real-time performance for target detection,and has a low false positive rate,which satisfies the actual situation application requirements.