Application Of Improved MRF Algorithm In Remote Sensing Image Segmentation Of High Consequence Area Of Oil Pipeline

With the development of remote sensing technology,remote sensing satellite can provide clear and better quality remote sensing image data,which contains rich ground feature details.Oil pipeline is a high-risk system for continuous transportation,and high consequence area refers to the area that would endanger the lives of the public,cause property loss,environmental pollution and other major damage.Therefore,the safety monitoring of the high consequence areas of oil pipeline is very necessary.The remote sensing image of oil fields are obtained by aerial photography or satellite,which can confirm the location of oil pipeline and the surrounding environment changes.When problems occur,corresponding safety measures can be taken as soon as possible.However,the monitoring of high consequence areas along the pipeline needs a lot of manpower and financial resources.At the same time,the realization of image segmentation in the high consequence areas of oil pipelines also has a very practical value.The Markov model is widely used in image segmentation because of its complete mathematical theory and good ability to express image spatial information.This paper mainly describes the image modeling.The main work of this paper is as follows:Firstly,a method of adaptively selecting potential function parameters for different pictures is designed to solve the problem that the Markov model only selects one potential function parameter for the entire image,and lacks the adaptability associated with the actual image,resulting in insufficient segmentation results.In this method,the pixels with similar or identical pixels are classified into one class according to the differences in gray levels,and pixels with similar properties are assigned the same potential function parameter.On this basis,a fuzzy potential function model that incorporates image pixel gray information,neighborhood correlation,and edge gradient information is proposed to obtain more subtle differences between pixels to describe the possibility of pixels being classified into the same class.In this way,the accuracy of segmentation is improved.Secondly,in order to solve the problem that the feature field and the label field contribute the same weight to the target energy function in the MRF MRF model,affect the segmentation result,and the fixed weight cannot achieve the effect of balancing the two parts.The constraint relationship between the characteristic fields,a variable so as to optimize the objective function of the algorithm.Thirdly,it is difficult to get the global optimal solution and the segmentation result is too random in the Markov algorithm based on ICM.By introducing the artificial bee colony strategy with accurate optimization and good robustness,the global minimum image energy can be obtained and the image segmentation result can be output.In order to enhance the timeliness of the algorithm,the pixel points in the image are divided into fixed points and fuzzy points.Thus,the efficiency is improved and the operation time is shortened.Finally,by visually observing the segmentation result map and calculating and comparing the peak signal-to-noise ratio(PSNR)and structural similarity(SSIM),it is proved that the improved MRF algorithm has better segmentation effect than other segmentation algorithms,and it is applied to the remote sensing image segmentation of high consequence area of oil pipeline to test the effect.