Narrow-band Radar Vehicles Classification Method And Implementation

With the development of radar technology, radar automatic target recognition has become the future direction of radar. Narrowband radar has a far detection distance, and it is widely equipment in China, so it will have great significance if the classification of narrowband radar targets can be achieved. However, the narrow-band radar resolution is low; it is difficult to obtain detailed properties of the targets. On the other hand, as wheeled vehicles and tracked vehicles are different in structure, there are large differences in micro-Doppler, so micro-Doppler information of the targets can be used in vehicles classification. This paper study on the narrowband radar moving vehicles classification method and implementation, the main content can be summarized in the following three areas:1. Micro-Doppler effect of moving vehicles are introduced, and the micro-Doppler model of vehicles is studied. First, the single scattering point for rotating micro-movement were introduced, the mathematical expression of its micro-Doppler signal model are given On this basis, the micro-motion model of the rotating body are given, and the related variables which affect the target micro-Doppler signals are expounded. Then the wheel and track micro-Doppler model are studied and the differences in mathematical expressions are deduced. Measured by the target vehicle radar echo signals are analyzed, the differences between wheeled vehicles and tracked vehicles echo has been pointed out.2. When introduce the classification process of moving vehicles targets, some common signal processing methods and classification methods were been studied. The target echo clutter suppression methods such as MTI, CLEAN, and GMF has been introduced, and by comparison of the different methods of clutter suppression effect, the advantages and disadvantages of each method are described. In feature extraction of target recognition, the feature based on Doppler distribution and energy distribution has been introduced. By classifying the real targets, the various features of the classification results are given. In the selection of classification algorithms, three algorithms such as LDC, KNN, and SVM are studied. Through the analysis of the algorithms, the conditions of these algorithms are given.3. For the classification of narrowband radar moving vehicles, the hardware engineering based on DSP are introduced. On the basis of the classification methods introduced above, and considering the real-time needs of the engineering as well as target classification results, the appropriate method to design target classification system are selected, and the DSP hardware engineering are debugged. By solving the problem of debugging, and analysis the computation time and the classification results, the feasibility and reliability of the narrowband radar vehicle target recognition method were confirmed.