Ship Target Detection And Sample Making On SAR Images

The ship target surveillance monitoring in the ocean is a traditional defense mission for our country, having important applications in civil and military sectors. The Synthetic Aperture Radar (SAR) is a kind of high resolution imaging Radar, with all-time and all-weather work ability. The ship detection on SAR images is one of the important research content of target detection. The ship detection, In essence, Segment the ship target pixels from the sea clutter on SAR images, acquiring the type and features about ship targets based on recognition method.However, with the continuous development of SAR technology and the further requirements of the detection, a new set of propositions are generated. Based on pressing practical demand, this paper carried out the following research work:The first, this paper studied the background and meaning of the SAR ship detection, compared the domestic and foreign development present situations, and summarized the overseas development of ship detection from application Angle and the research status of domestic focusing algorithm. This paper analyzes the new problems and directions which should be solved urgently in ship detection.The second, this paper explores the imaging mechanism applied in ship detection in SAR images, analyzing the method of SAR images’ preprocessing for the practical requirements in ship detection, such as noise filtering, incidence angle correction and geometry correction. This paper also analyzes the factors of the SAR system and sea state influencing the ship detection, etc.The third, this paper sums the sea clutter statistical model on the False Alarm Rate (CFAR) ship detection method, studied the different statistical model of formula derivation, parameter selection and the applicable conditions, and compared the advantages and disadvantages of probability model, etc. And on this basis, this paper introduces studying more multistage CFAR algorithms.The last one, in view of the present domestic ship recognition direction the theory analysis is more, most based on the artificial experience and simulation datum.In order to analyze SAR ship target detection algorithm performance and conduct further research, the paper introduces two experiments in detail, one based on the analysis of the field of ship information and the other based on SAR image and AIS jointing detection.Based on the above discussion of study, around the ship detection related problems which are urgently need to solve, some more innovative work were carried out in this paper, as follows:(1) For the high noise single-look complex SAR images a Rice distribution CFAR detection algorithm is studied, which improves the detection precision for the traditional probability model under the condition of large noise, and analyzed the experimental results comparing the classical model of comparative analysis.(2) On analyzing the characteristics of ship targets in SAR image, and considering the false alarm rate parameters in CFAR algorithm having too much dependences on artificial prior knowledge which makes the final detection result is unstable. This paper based on the SAR target high brightness, high density studied the ship characteristic detection algorithm, and comparing with traditional morphological processing methods, the advantages are more obvious.(3) Detailing the construction of ship sample library based on high resolution SAR images, and ship target geometric feature extraction methods are studied. Based on the analysis of the polarization and incident Angle influencing ship length. The above study made certain foundation to the SAR images ship identification in the future.