Research On The Key Technologies For Ship Surveillance By Integration Of Space-borne SAR And AIS

With the increasing demands for ocean surveillance, more and more countries take efforts to develop the surveillance technologies. With the prominent advantages of all-day-and-night, all-weather and wide swaths working capabilities, space-borne Synthetic Aperture Radar(SAR) has been widely used in the maritime surveillance. However, ship classification using space-borne SAR image is still an open issue, which affects the efficiency of ship surveillance greatly. To satisfy the urgent maritime surveillance demands, to deal with the difficult technical problems, to focus on the developments of ocean surveillance and to investigate based on the 12 th Five-Year pre-research projects, this paper researches on the key technologies for ship surveillance by integration of space-borne SAR and AIS, in accordance with the fact that Automatic Identification System(AIS) could aid to improve the ship interpretation performances using SAR images.Ship surveillance by the integration of space-borne SAR and AIS is an information fusion problem essentially. On the basis of the summary and analysis of its research background, status quo and development, the key technologies for ship surveillance by integration of space-borne SAR and AIS are correctly derived. Considering the data fusion on feature layer, this paper emphasizes on investigating the key technologies including the pre-processing of space-borne SAR and AIS data for fusion, data association of space-borne SAR and AIS, and the fusion detection and identification of ships using space-borne SAR and AIS information.The basis of this research is the pre-processing of space-borne SAR and AIS data for fusion. To derive ship’s feature information accurately, feature extraction methods based on space-borne SAR images and AIS report decoding are investigated in detail. To solve the problems for ship detection under complex sea conditions using SAR image, a novel adaptive Constant False Alarm(CFAR) detection method is proposed to deal with the high false alarm problem under the non-homogenieous sea conditions. The extraction methods for ship’s geometric parameters, geographical parameters and motion parameters are summarized. Also, the AIS information decoding and checking methods are investigated and implemented using the reports of space-based AIS “TianTuo I” and shore-based AIS. Meanwhile, on consideration of the quality and error level of SAR sensors’ data comprehensively, the time and space calibration techniques are investigated using space-borne SAR and AIS data.The kernel of this research is the data association of space-borne SAR and AIS. Firstly, data association based on position information is investigated. On the basis of analysis of various important factors affecting the association, its three key technologies including position projection, position prediction, and searching and matching are investigated in detail and validated. For the purpose of enhancing the validity and improving the performance of data association based on position feature, three key technologies are improved respectively. To deal with the high-sea-state and non-linear motion conditions, improved ship movement prediction models are presented, including position prediction model under varying course condition and the position gray prediction model. Due to the Doppler displacements for the moving ships in the space-borne SAR image, improved methods for the estimations of projected course and speed are proposed on the basis of analysis of main factors and error level of Doppler displacement. For the conditions that there exist larger systematic errors and the high-density-shipping condition, data association of space-borne SAR and AIS based on the topology structural characteristics-Coherent Point Drift(CPD) based association is investigated. Its performances are also validated. On the basis of the research for data association based on position information, the data association based on position and attribute features’ information fusion is also explored. The association method is proposed, and three key technologies including feature selection,fusion decision and decision evaluation are investigated. The characteristics of four fusion decision methods including weighted average, non-linear multiplication, Dempster-Shafer(DS) evidence theory reasoning and fuzzy comprehensive judgement are compared and analyzed.The purpose of this research is the fusion detection and fusion identification of ships using space-borne SAR and AIS information. Compared to the ship detection and classification using space-borne SAR image, it could improve the efficiency for ship surveillance. The model of fusion detection and identification of ships based on space-borne SAR and AIS information is proposed, which provides feasible schemes for fusion detection and identification. Especially, a novel ship classification method based on Analytic Hierarchy Process(AHP) using high resolution space-borne SAR imagery is also proposed and validated, which provides valid evaluation criteria for feature selection and classification decision, and benefits the research on fusion identification. The experiments for fusion detection and identification are finally implemented, and the results validate the feasibility of the schemes for fusion detection and identification by integration space-borne SAR and AIS.