Sea Ice Drift Monitoring Technology Research GOCI Satellite Based On Image Processing

Sea ice disaster is one of the major ocean natural disasters in Yellow Sea and Bohai Sea, and it is an important marine environmental condition affecting the ship transportation, drilling construction and production operation in ice area. It also has been one of the important environmental factors must be considered in shipping, development and production in the area of Yellow Sea and Bohai Sea. Sea ice drift monitoring is an effective way to improve the sea ice disaster forecast accuracy and reduce sea ice disaster. However, due to constraints of environmental conditions, time and space, it is too difficult to implement and operate a wide range of sea ice monitoring by traditional methods. Remote sensing images acquired from Korean GOCI satellite is the primary basis for the wide range sea ice drift monitoring in Bohai Sea and Yellow Sea at present. But these satellite images are still mainly rely on manual ice drift image interpretation, and thus it needs the research of sea ice drift automatic monitoring technology badly to improve the image interpretation efficiency and promote the development of sea ice drift business monitoring.According to the actual business needs of the North Branch Prediction Center, this thesis researches the large-area sea ice drift monitoring technology in GOCI satellite images, and proposes a sea ice drift automatic monitoring algorithm by utilizing morphology, cluster analysis, SIFT feature matching and rule-based reasoning methods.The main research contents of this thesis include the following aspects:(1) The relevant theoretical knowledge, such as image matching and scale-invariant feature has been studied. This thesis focuses on the design principle and the basic idea of SIFT algorithm and applies it to the sea ice feature matching in GOCI satellite image.(2) By detailed research on image segmentation, morphology, clustering and other image processing methods, this thesis designs a sea ice recognition algorithm of GOCI satellite images, which can automatically identify the different sea ices presenting in the image.(3) For the problem that SIFT algorithm cannot match the sea ices occurred deformation with time, this thesis proposes the sea ice matching inference rules based on expert experience, which can match the sea ices occurred deformation with time by combining the results of the sea ice identification and SIFT feature matching.(4) According to the business needs of the North Sea Forecasting Center, this thesis develops a satellite sea ice drift monitoring system, which can automatically calculate the sea ice drift direction and distance according to the two satellite images.