Research On Ship Wakes Detection In SAR Images

The main work of the dissertation is one of the key aspects in SAR oceanic applications, namely, ship wakes detection in SAR images. The dissertation focuses on two aspects, that is, research on ship wakes detection in SAR images, and Field Programmable Gate Array (FPGA) implementation for one detection algorithm in SAR images.Firstly, the dissertation introduces the basic theory of SAR image Interpretation and Understanding, and Automated Target Recognition (ATR) . After briefly listing the existing major detection methods, I summarize the mechanism of detection algorithms in SAR images and compare every method with others in the aspect of its limitation and the processed effect. Generally speaking, there're two sorts of detection methods, one based on the geometric characteristics of SAR wakes and the other based on the statistical differences between areas containing SAR wakes and areas containing only ocean background. At last, I mainly introduce two detection algorithms in the frequency domain. One is based on the differences in statistical characteristics between areas containing SAR wakes and areas containing ocean background. The other is used in some specific situations when SAR images represent some periodic waves. After discussing the theoretical feasibility, the effectiveness of this algorithm is demonstrated by the test results of simulated wake images and actual SAR images of ship wakes. The results indicate that these two algorithms can detect some weak wakes in SAR images effectively.In recent years, the applications of ATR system in SAR oceanic field are developed increasingly. FPGA implementation is used widely in image processing. Therefore, how to use FPGA in SAR oceanic field is worth to discuss. In this dissertation, the other key issue is the groping research on FPGA-based architecture for the detection algorithms of moving ships in SAR images.Firstly, I compare two hardware implementations for SAR image detection system and make the conclusion that FPGA is most suitable. Then, I introduce the FPGA implementation for the existing algorithm ~ using flatness and the dissymmetry characteristics of the distribution shape in detection algorithms. The proposed architecture consists of data interface, software module and hardware module. The hardware is implemented on Altera StratixII ep2s30 FPGA using VHDL. It accounts for the main complex calculations. The software retrieves the vectors calculated in hardware and performs complementary computations. The design has proved to be useful for future SAR image detection system.At last, I introduce some improving proposals to accelerate our processing speed such as improving the way to use sliding windows. I also discuss the problems in hardware design and other problems concerning with future SAR image detection system.