Research On Application Of Multi-modal Remote Sensing Image In Railway Construction Supervision

With the rapid development of railway construction in China,the construction period of railway lines is long,the span is wide,and the geological environment is complex.In the process of construction,problems such as illegal occupation of land and ecological damage will inevitably occur.The traditional ground survey method is single,and it is difficult to truly reflect the construction situation of the whole line.Satellite remote sensing has the characteristics of wide detection range,short reentry cycle and low cost,which can effectively make up for the deficiency of ground survey.Because the single mode remote sensing data is sometimes missing,it will cause other problems such as discontinuity of time series.The comprehensive analysis of multi-mode image can solve many problems that single mode image cannot solve.The research mainly uses multimodal remote sensing images to monitor the changes of the railway and its surrounding areas during the railway construction cycle to provide data support for auxiliary railway supervision.The research contents mainly include the following aspects:(1)Research on multimodal remote sensing imaging registration and image fusion based on superpoliographic matching.Improved Space Constraint Based on Space Constrained Mixed Model and Image Matching Strategy,Space Constraints Based on Markf Flower(MRF),introduced into the Canny edge extractor to achieve a good boundary Adhesion and intensity uniformity,alleviate the problem of superpoxy segmentation on bright changes,noise,and blur boundaries.Second,ultra-pixel-based image matching is achieved by improving key hyperpox detector and ultra-pixel space pyramid.Unlike the traditional detection based on key points,the two images are directly matched through superpoxy to superpixes.The advantage of this method is that the matching requirements of key hyperpoxes and fewer local features are small,and the method greatly reduces the calculation requirements,which has very high tolerance in the noise in the observation data.(2)Research on the detection method of remote sensing image of the airport based on superpoxy conditions.First,the multi-mode remote sensing algorithm is divided into homophysical ultra-pixes,and then calculate the characteristic difference information between the corresponding ultra-pixel block to obtain a variation image,and finally construct a conditional random field on the ultra-pixel level.Change detection model.This method can comprehensively consider the spectral,texture,and space contextual information,which can better maintain a variable region boundary detail,reduce a large number of noise generated by spectral change sensitivity and registration error,reduce changes in change detection,and improve change detection.Accuracy and accuracy.(3)Acquisition and evaluation of railway construction supervision information.Using the above algorithm to varying the railway area to detect,the following types of data are obtained.First,use the time series image to evaluate the construction progress of railway engineering.The second is to monitor assessment of the monitoring assessment of the mixing station,construction camp,construction area,and abandoned slag field,and to master whether it is compliant,and evaluates the safety status.Third,environmental monitoring is carried out on the recovery of construction sites after the completion of the railway,and vegetation coverage is carried out to see whether there is an ecological recovery standard after railway construction.(4)Design and Implementation of Multimodal Remote Sensing Image Railway Construction Supervision System.The secondary development of the open source platform will be packaged,including the applied algorithm,including data pretreatment,ground object classification,information extraction,multimode image change detection,rail construction progress assessment,temporary engineering footprint and vegetation coverage recovery testing,The topic figure generates functional modules such as report output,and the existing algorithm is packaged with existing algorithms,and the integration of plug-ins will be achieved,and finally realize the convenient and fully functional and fully functional railway construction supervision system.