Research On Technology And Detection Probability Of Min - Min Compound Detector

In this dissertation, a research has been made on the signal outputting, processing, feature extraction and integration, as well as the reliability in the background noise of tri-mode composite detector (MMW radiometer, long IR detector, LFMCW radar) in use of terminal sensing detectors.Firstly, on the combination of radiation characteristics of the tank and background, the output signal models are established of each detectors in case of conical scan. According to the theory of scattering centers, a one-dimensional HRRP is simulated of a typical tank, and factors are studied.Secondly, scanning track and detection zone on the ground are studied during stable scanning. Dynamic output models of each detectors are established. Preprocessing and signal extraction method of one-dimensional HRRP are also studied.Thirdly, the information fusion of the composite detector is analyzed from the feature level, and support vector machine (SVM) is selected as the algorithm. An improved One-Class SVM is proposed and the effectiveness of the improved algorithm is verified.Forthly, the power, amplitude probability distribution and SNR-improving methods are analyzed. LFMCW radar clutter model is established using the γ-f model. The false alarm and detection probabilities of different SNR are compared using statistical methods.Finally, according to the design requirements, hardware is designed, assembled, debugged. Outfield experiment is made to collect experimental data of simulated tank. The effect of the improved SVM algorithm is verified based on the data.