Variational Waterline Extraction Method And Application Based On Remote Sensing Data In Xiamen Waters

Waterline refers to the instantaneous transition line between the constantly fluctuating ocean water and land recorded by remote sensing.As an important and fundamental step in mapping coastal zone,monitoring natural disasters,and studying the dynamic change and trend,waterline can be used to monitor the floodplain,study the topography of the inter-tidal zone,map shoreline,detect reef and underwater islands,and analyze the coastal sediment deposition.Waterline analytical techniques and applications are of great significance in marine scientific research,production activities and national defense construction.In this paper,optical remotely sensed imagery was used as data source.Waterline extraction model incorporating both spectral features and edge features was established to address the difficulty of waterline segmentation with the active contour in the case of the weak boundary.Furthermore,by introducing the prior knowledge into the variational framework,the improved model was proposed to extract the waterline in the condition of complex weather such as a heavy cloud occlusion.Waterline was used to solve the problem of lack of topographic data in the inter-tidal zone of Xiamen and establish the tidal floodplain numerical model.The main research and achievements of this thesis are as follows:(1)By integrating spectral and edge features,the waterline extraction problem in the condition of weak boundary was studied.Taking geodesic active contour as the main research object,the limitation of the application in waterline extraction was analyzed.The waterline extraction model was established by extracting the edge features and spectral features of water bodies.The model retains the advantages of sub-pixel precision of geodesic active contour used to extract waterline,which improve the accuracy and efficiency of waterline extraction in the condition of clear land boundary,and effectively solve the problem of the active contour leaking at the weak edge.(2)The waterline extraction model with prior knowledge constraint was used to deal with the difficulty of waterline extraction in the condition of complex weather and cloud occlusion.The waterline extraction model with prior knowledge constraint was presented by using the historical waterline contours as the prior shape,collecting the historical waterlines as the training samples,defining and introducing the prior shape constraint into the active contour model.In the process of contour evolution,the model measured the similarity between the evolution curve and the prior shape.The waterline partially covered by cloud was extracted by curve evolution under the influence of prior shape constraints,spectral features,edge features and prior shapes.(3)The waterline was applied in the numerical prediction of offshore marine hydrology.The problem of topographic acquisition of inter-tidal zone was studied.The temporal and spatial characteristic of tidal flats was analyzed and the remote sensing technology of tidal flat topography was discussed.A series of methods were compared and analyzed from different evaluation indexes.The results indicated that the method of remote sensing waterline is an effective method to measure tidal flat topography.This method was used to extract multi-temporal waterlines and calculate the topography of inter-tidal zone in Xiamen,which made up the gap of inter-tidal terrain data of sea chart.These inter-tidal terrain data and sea chart data were used to generate high-resolution grid data required by numerical calculation of tidal floodplain in Xiamen Sea.The numerical model of tidal floodplain in Xiamen Sea was established and used for forecast verification.