Target Detection And Discrimination In UWB-SAR Image

To detect concealed targets in foliage with Ultra Wide Band Synthetic Aperture Radar (UWB-SAR), the problem of target detection and discrimination in UWB-SAR image is systematically studied in this thesis.At first, the present research state, methods and significance of UWB-SAR target detection and discrimination are summarized, and the situation of UWB-SAR target detection and discrimination in UWB-SAR automatic target recognition system and its current existing problem is also pointed out.The problem of Constant False Alarm Rate (CFAR) target detection in UWB-SAR image is studied in chapter 2. The clutter statistical property of different vegetation bestrow is analysed, and a CFAR target detection method is proposed based on the corresponding conclusions. Under a common form, the method can perform CFAR processing in different vegetation bestrow clutter. A fast statistics computation method is proposed for CFAR processing, which can save the computation expense a lot.In Chapter 3, the target feature extraction method is studied for target detection and discrimination in UWB-SAR image. Based on the UWB-SAR echo model and image principal, the target feature extraction method from aspect domain, frequency domain and resolution domain in UWB-SAR image by filtering in frequency domain is presented, and the corresponding filter design are offered. With the constructed UWB-SAR target model, some statistical analysis methods and Hidden Markov Model are modified and combined for feature fusion. The point target detection method in UWB-SAR image is studied, and the result is used for trunk clutter suppression. The effectiveness of these methods is confirmed by the experiment results.The problem of nonhomogeneous environment target detection in UWB-SAR imagery is studied in chapter 4. To conquer the clutter model excursion with the change of vegetation bestrow, based on the modification of a truncation clutter recognition method, a CFAR target detection method that can adapt to the change of clutter statistical model is presented. To deal with the clutter edge at the boundary between a block of intensive clutter and a block of weak clutter, based on the variance index, a CFAR target detection method is proposed. The experiment results show that the methods are effective.In chapter 5, the quantificational performance evaluation method of UWB-SAR target detection and discrimination is studied. Based on the receiver operating characteristic curve, two methods are presented for quantificational performance evaluation of target detection. Using the separability of the target and clutter as the measure, two indexes for evaluating the performance of target feature extraction are introduced. A calculation method of several parameters for UWB-SAR target detection and discrimination is modified to cope with the speciality of UWB-SAR target and clutter.At last, the research of the thesis is summarized and some problems and interesting area for future research are pointed out.