Study Of Radar High Resolution Range Profile Target Recognition

With the development of modern war informationization degree, detection and ranging function of the conventional radar is not enough to the requirements of modern war about real-time dynamic monitoring and processing. Thus radar target recognition has become one of the important development directions of the modern radar system. Radar high-resolution range profile(HRRP) can provide the distribution of target scattering points along the radar line of sight direction, reflecting the structure and shape information of the target, and is relatively easy to obtain compare with synthetic aperture radar profile. It is one of the hotspots in the field of radar target recognition.In this paper, combined with the Radar HRRP target recognition frame of theory and engineering application conditions, carries out the research on data preprocessing, template library building and online identification problems, summed up the main work completed is as follows:1. Surrounding the basic structure of the HRRP target recognition system, in this paper, commonly used high resolution range profile modeling method is introduced, image model of HRRP is given, and sensitive characteristics of HRRP is analyzed, the ideas and methods of feature extraction is introduced, and commonly used recognition method is also discussed. Through this basic principle of the Radar HRRP target recognition is described.2. The second part focuses on data preprocessing problem. Abnormal track is eliminating, extraction method of target echo area from coarse to precise is given, the advantages and disadvantages of three kinds of commonly used distance alignment method is compared and function and effect of non-coherent average and normalizing echo amplitude is discussed.3. Based on analysis of the characteristics of non-cooperative target data, a novel recognition method of template update and fusion. Firstly, a single route data is framed and model in order to complete the basic template generation. An improved Jensen-Shannon divergence(IJSD) is given as similarity measurement. With IJSD, template update algorithm based on single route data fusion is given. this method can effectively optimize the parameter of statistical distribution of each sub frame. Analyzing the characteristics of template library, Bayes classifier is used for object classification. The experimental results verify the effectiveness of the proposed method.4. Consider the problem of the continuous observed same target may not be classified into the same result in real recognition, an online target recognition method is proposed. The core idea of the proposed method is determining the target category according to multiple recognition results of the same target. Cumulative confidence is the basis for judgment. Calculation method of single recognition confidence is discussed. Two kinds of cumulative confidence calculation is given. Experiments show that, the uses of multiple recognition results accumulation can effectively processing the encountered problems, and to some extent improving the target recognition rate.