ScanSAR Mode Sea Ice Image Segmentation

Synthetic aperture radar(SAR)provides an efficient imaging tool for sea icemonitoring,which has found important applications in both scientific andoperational activities such as climatic research and ship navigation. Sea ice hassignificant scientific value in global climate research as a sensitive indicator ofglobal climate variability,and also has great strategic value in global economypattern transformation has been gradually enhancing，which contain the NorthwestPassage and the Northern Route. Synthetic aperture radar images are now playing akey role in sea ice studying. SAR sea ice image segmentation is an important partof the image interpretation. The backscattering coefficient of object is changed dueto the variation of incidence angle of SAR electromagnetic wave beam, whichcauses the intensity inhomogeneity along SAR range direction in SAR sea iceimage.That produced serious influence on the accurate segmentation of SAR sea iceimage.For a wide range of SAR sea ice image, especially in wide swath spaceborneand airborne SAR sea ice image interpretation, the angle incidence effect can not beignored. The main contents and contributions are organized as follows:Firstly, the paper provides an overview of the related background about SARimage segmention, the angle incidence effect and the researches at home andabroad.We provides relative concepts.Then we generalized and classified theexisting relative research literature.Review and summarize the SAR imaging,theprocess of the development of SAR and the principle of SAR(includingfeature,SAR speckle formation mechanism,incidence angle,multilooking et al). Theconcept of image segmentation is introduced.Then introduce SAR imagesegmentation methods such as threshold segmentation, texture segmentation andedge detection.Markov Random Field(MRF) segmentation method is key of imagesegmentation.Secondly, the traditional mothods only consider the SAR image of localnonstationarity, while the incidence angle effect of such large scale changes can notguarantee the wide observation with spaceborne and airborne SAR sea ice imagesegmentation accuracy. Incidence angle effect is a function of the sea ice type. Thetype of correction is the premise of sea ice of which types are known and segmentation isessential to low level classification.This is in the segmentation process integrated incident angle correction step that may be provided. Based on wide observing band data,thispaper puts forward a kind of integrated incidence angle effect correction steps ofthe segmentation algorithm,which considers the speckle noise, the incidence angleeffect and uncertainties. Sea ice regions are reserched as objects,which doesregion-based intial clustering,then does type of correction-based incidence angleeffect and region merging to get the final segmentation one by one.Finally,we summarized our reaserch work for the thesis,and discussed someresearch topics and directions relative to this work in the future.The algorithm is applied to Radarsat-1satellite ScanSAR Wide mode andScanSAR Narrow mode SAR sea ice image. The experimental results demonstratedthe efficiency of the proposed method．...