Ship Target Detection In SAR Imagery Of Land-water Conjunction Area

Recent years, the detection of ships in SAR images of land-water conjunction areas,e.g. harbors, islands, has drawn considerable attentions. Obviously, it has great potentialin both civil and military applications. However, the extreme heterogeneousenvironment of such areas, as well as the existence of varying kinds of static andmoving targets within and around, leads to many intractable problems for ship detection.Therefore, this paper focuses on the problem of ship detection in land-water conjunctionareas by the usage of SAR images. After the imaging process of colleted SAR data ofland-water conjunction areas, non-parametric CFAR methods for detecting ships inSAR imageries of such areas have been deeply researched. Besides, moving shipdetection methods using single-channel SAR imageries of such areas have also beendeeply investigated. The main contents of this paper can be summarized as follows:Chapter1introduces the state of the art and trends of the techniques for detectingships in SAR imageries of land-water conjunction areas as well as the background andsignificance of the subject, and summarizes the main work done in this paper.Chapter2studies the problem of SAR data collection and SAR imaging. Avehicle-borne SAR experimental system used in data collection is introduced in the firstplace, the applicability of which is also addressed. Then, the selection of high resolutionSAR imaging algorithm for the colleted data is discussed based on the parameters of theexperimental system and the restrictions of the experiment conditions. Finally, SARimaging procedure on the collected data of land-water conjunction areas is implementedusing a proper R-D algorithm, and the background complexity of the SAR images andthe difficulties of ship detection are also well discussed.Chapter3studies the methods of non-parametric CFAR ship detection in the SARimageries of land-water conjunction areas. First, classic parametric and non-parametricmodels for SAR imagery background modeling are introduced. Then, a newnon-parametric statistic model based on finite series of orthogonal polynomial isproposed, considering the extreme heterogeneity of the SAR imagery backgrounds ofland-water conjunction areas. Analysis results show that this new model can fit thecollected background data of land-water conjunction areas quite well in comparisonwith classic models. Consequently, a CFAR detector based on the orthogonalpolynomial model is proposed. This detector has a simple structure and a goodperformance. The effectiveness of the detector is verified by collected field data.Chapter4studies on the methods of moving ship detection using single-channelSAR imageries of land-water conjunction areas. For low-cost small SAR systems,moving target detection using only single-channel SAR images has great significance inapplication. First, the SAR echo model of moving targets and the effect of target motion in SAR images are analyzed. Then, with the collected single-channel SAR images ofland-water conjunction areas, the frequency-domain filtering method and the subaperture canceling method are applied in the research of moving ship detection. Thedetection results are analyzed subsequently.Systematic conclusions are finally drawn in Chapter5, as well as the defects of thispaper and the prospects of further research.