| As its well synthetical performance, millimeter wave detecting technology has become one of the most current developing directions of precise detection. Along with the development of detecting technology and the complexity of spotting background, more and more information of the target is needed to be required, not only the distance, speed and angle of target, but also the accurate tracking and localizing to the target. Therefore, the compounding detection is an inevitable trend of development. Research of this dissertation is just based on millimeter wave active/passive compounding short-range detecting technology. Active detection adopts MMW short-range high-resolution radar, which has far detecting distance and can access the speed, angle, distance and other information of target. Passive detection uses MMW radiometer, which does not emit electromagnetic wave, has no glitter effect and works snugly. The compounded detecting system, working together by these two systems, can make the advantage to complement each other, so to obtain the more detailed information of target and the higher rate of target identification.Signal processing is an indispensable part of a precise detection system. Advanced signal processing technology is a primary factor to improve the accuracy of detecting system. So the dissertation is focus on the thorough analysis and research of target identification system of MMW active/passive compounding detection. For active detection, through the different methods of waveform design, ways to synthesize high-resolution range profiles by different high-resolution signals is researched. And then, based on matching pursuit time-frequency analyzing method and appropriate atomic library, the echo signal is sparse decomposed and restructured to extract the features of the signal. At last, by the way of relevance vector machine and fuzzy relevance vector machine, target identification of the MMW short-range high-resolution radar range profile is realized. In passive detection, the radiation signal of target is denoised by the method of sparse decomposition at first. Then, based on the features of waveform in time domain, frequency domain, and time-frequency domain, target identification by application of rough set theory and artificial neural networks is detailed. The rough neural networks, organically combined of the two methods, may accept the data set with up-and-low boundary, which improves the performance of network and displays enormous superiority in information processing.Following above, target identification of active/passive compounded detecting data fusion is discussed. Based on the architecture and the level of information fusion theory, a spatial-temporal fusion structure of target feature is designed, according to the characteristic of MMW active/passive detecting systems. Method of D-S evidence theory based on fuzzy clustering is adopted in space domain fusion for active and passive detecting signal. And then, the space domain fusing results of different heights are taken as the measurable functions, using fuzzy integral approach to secondary time domain fusing. So that, a more reliable consistent description of the target can be achieved, and the fusion results of target identification is improved greatly.Finally, the realization of target identification technology in MMW active/passive short-range detection system has been designed in the dissertation. Due to the complexity of working mode, large amount of information processing, and high requirement of real-time processing, high-speed chip DSP and FPGA have been chosen to construct a signal processing system. The design may satisfy the requirement of speed and flexibility synchronously by combining the advantages of DSP and FPGA. The software and hardware design of signal processing system is work out and correlative experiments have been down. Furthermore, based on the actual circuit, signal integrity in high-speed digital circuit design is analyzed to achieve better circuit design.
|
PDF Full Text Request