Remote Sensing Detection Method Research For Environment Hidden Dangers Elements Of Complex Structural Along High Speed Railway

With the rapid development of high-speed railway,the operation safety of highspeed railway has also been widely concerned.Geographical environment along highspeed railway is complex,and there are a lot of potential environmental safety hazards.According to the "Safety Protection Management Measures for High-speed Railway(Draft for Opinions)",the behavior of quarrying,mining,sand excavation and soil extraction in the safety protection zone of high-speed railway should comply with the national standards,industry standards and railway safety protection requirements,which improper behavior may cause the collapse or sedimentation of high-speed railway roadbed,and it seriously threaten the operation safe of high-speed railway.In this paper,the surface damage area caused by quarrying,mining,sand excavation and soil extraction is called excavation area.Therefore,the investigation of the excavation area in the high-speed railway safety protection zone can effectively detect the behavior of quarrying,mining,sand excavation and soil extraction.However,the coverage of high-speed railway is becoming more and more widely.The traditional methods,such as manual field investigation or ground sensor monitoring,are time-consuming,inefficient,poor timeliness and difficult to form effective real-time monitoring for the whole high-speed railway safety protection area.High-resolution remote sensing technology has the advantages of macroscopic,periodic,real-time and comprehensive.It's a very effective technical means to monitor the environment hidden dangers elements of complex structure in high-speed railway safety protection areas.The paper chooses the area of 5 km along Kunming-Kaili Section of ShanghaiKunming High-speed Railway as the research area,which are from the 19 th level image of Google Earth.In the framework of excavation area detection,firstly,the sample database of excavation area is constructed.Then,construct a multi-scale convolution neural network framework(MCNN)by improving the traditional CNN.Finally,the candidate excavation areas extracted by saliency detection and connectivity analysis are used as the input of the trained MCNN model to achieve rapid and accurate detection of excavation areas.After experimental analysis,the main conclusions are as follows:1)In this paper,the surface damage area caused by quarrying,mining,sand excavation and soil extraction in the safety protection zone along the high-speed rail line is called excavation area.And the sample database of excavation area is constructed by combining AID remote sensing data set and excavation data set.Using data enhancement improves the diversity and generalization ability of the sample database.2)In this paper,a detection framework of extraction area is constructed,which includes two modules: extraction of extraction candidate areas and detection framework of extraction area based on MCNN.Compared with the direct input of the original image,the processing flow in this paper can reduce the input data of MCNN by 96.47%,and greatly improve the detection efficiency of the extraction area.The MCNN framework in this paper can receive input image of any scale,it realizes the detection of extraction area about 600 km from Kunming to Kaili section of Shanghai-Kunming high-speed railway.And quantitative analysis of the detection results of extraction areas in Pingba District of Guiding and Anshun City shows that the correct rate is about 54.05%,the recall rate is 95.24%,and the total IOU is 0.73.