Research On Recognition Method Of High Consequence Areas Of Oil Pipeline Based On Remote Sensing Image

Oil pipeline is a high-risk system for continuous transportation.It is of great significance to ensure its safe operation and to provide early warning of possible leakage accidents.The high consequence areas(HCAs)of a pipeline are defined as the area where the leakage of a pipeline may cause greater adverse effects to the public and the environment.The identification of high consequence areas includes two steps: determining the type of ground features along the pipeline;Measure the distance between the pipeline and specific features to delineate the areas that may be affected.The oil pipeline has a long mileage and many changes in the surrounding conditions along the line.The traditional methods of manual line inspection and field measurement are difficult to adapt to the increasingly stringent management requirements.In this paper,the method of identifying high consequence areas of oil pipelines based on remote sensing images is studied.The remote sensing image classification algorithm is used in this method to obtain the feature information along the pipeline,and modeled in the geographic information system(GIS)together with the pipeline information.According to the recognition criteria of high consequence area,the spatial analysis technology in GIS is used to calculate the topological relationship between oil pipelines and ground objects,and to identify HCAs.The main work of this paper is as follows:Firstly,an improved remote sensing image segmentation method based on manifoldsimple linear iterative clustering algorithm is proposed to solve the problem of low accuracy of remote sensing image segmentation by simple linear iterative clustering algorithm.The improved method incorporates multi-segment spectral features and texture features to perform content sensitive superpixels segmentation of remote sensing images,which overcomes the problem that the SLIC algorithm cannot be directly used in multi-spectral remote sensing images and does not make full use of the image feature information and the fixed size and number of super-pixels,resulting in low segmentation accuracy.Secondly,to improve the classification accuracy of the graph based semi supervised classification method,the fusion of object-oriented remote sensing image classification idea is used to improve the graph based semi supervised classification method.The proposed method uses superpixels to replace the original pixels in remote sensing images to reduce the scale of the graph.Using superpixels as the reference area to extract features,a variety of distance measurement methods are used to calculate the edge weights between superpixel nodes,and a graph structure that accurately represents the relationship between data is constructed.Finally,the proposed classification method is applied to high-resolution multi-spectral remote sensing images along the oil pipeline,using a small amount of labeling data to obtain the high consequence areas features information from the image.Then,combined with GIS,the actual situation of the pipeline was modeled based on the target contour coordinates of the high consequence areas and the pipeline coordinates obtained.According to the identification criteria,the topological relationship between the pipeline and the ground object is calculated by using the spatial analysis tools in the geographic information system,and the high consequence area of the oil pipeline is identified.The identification method is used to design and implement the identification system of high consequence area of oil pipeline based on remote sensing image,and the effectiveness of the system and identification method is verified by comparing with the actual situation.