Research On Feature Extraction And Parameters Estimation For Radar Targets With Micro-motions

The term"micro-motion"is defined as the mechanical vibration or rotation of a target or its components in addition to its bulk translation. The micro-motion status of a target can well reflect its sophisticated features. Therefore, it is of important value in military field, and draws great attentions to the researchers from the areas of target detection and recognition. In this dissertation, we keep our focus and do intensive research works on the topic of radar signature extraction from targets with micro-motions based on the modulation effects of the radar echo on this specific kind of targets.We introduce the background and significance of this research topic as well as the related works in Chapter 1. The state of the art is summarized under four aspects: the radar echo modeling of targets with micro-motions; the separation and feature extraction of micro-Doppler; the radar imaging of targets with micro-motions; and the target recognition based on micro-Doppler. We analyze the basic methodologies and techniques commonly used in this area, and point out their main problems. The major scientific contributions of this dissertation are summed up at the end of this chapter as well.In Chapter 2, we analyze the modulation effects of micro-motions on the radar echo of targets systematically. The modulation effects and characteristics are investigated according to micro-motions, the scattering structure of the target and radar waveforms. The radar echo model of targets with precession is established. The micro-Doppler and Doppler spectrum of targets with precession are deduced and analyzed. Under the two different conditions of the typical wideband waveform– LFM (Linear Frequency Modulation) signal and SF (Stepped Frequency) signal, we do deduction and analysis on the modulation regularities of the HRRP (high resolution range profile) with different parameters of radar waveforms and micro-motions respectively. The influences of target characteristics and radar waveforms on micro-Doppler are analyzed, including micro-motion pattern, scattering structure, radar frequency, instantaneous bandwidth, Doppler resolution and coherence of echos. The conclusion is verified by a series of experiments conducted both on the simulated and measured data.The methodology of analysis and extraction of micro-Doppler signatures are deeply studied in Chapter 3. We firstly analyze the performances of the different typical time-frequency distributions in micro-Doppler extraction. As the ballistic targets have both the high-velocity bulk translation and micro-motion characteristics, we propose an approach to estimate the bulk velocity based on a polynomial model of instantaneous frequency. With the support of this approach, the micro-Doppler and Doppler spectrum of the bulk motion can be separated. Then, aiming at the cyclostationary characteristic of the radar echo of targets with micro-motions, an estimation method of the micro-motion cycle based on CSD (Cyclic Spectral Denstiy) is proposed, and its performance is discussed afterwards. According to the characteristics of micro-Doppler on time-frequency distribution images, we estimate the precession angle by utilizing the inverse Radon transformation. While analyzing the bias of the estimation, we propose a method for correcting that of the precession angle.We elaborate the super-resolution estimation of micro-motion parameters in Chapter 4. Since the radar echo of targets with micro-motions is sparse, we establish a model based on sparse representation for estimating the micro-motion parameters, design the atom model of targets with micro-motions, and analyze the mutual coherence of the dictionary. For solving the problem of strong mutual coherence of the dictionary, we optimize the dictionary design by applying the theory of linear transformation. The mutual coherence of the dictionary can be effectively reduced with the solution's sparsity property held. In addition, we develop a sparse-representation-based algorithm for the micro-motion parameters estimation by integrating the dictionary transformation with the sparse solution solvers. The performance of the algorithm is verified as being acceptable by experiments. The radar echo of targets with micro-motions belongs to multi-component AM-FM signal models. Taking this characteristic into account, we propose a time-varying autoregressive (TVAR) representation model of the radar echo of targets with micro-motions. The methods of solving the TVAR model's parameters are discussed, and the micro-motion parameters are estimated.The research work and main innovative contributions of this dissertation are concluded in the final chapter. We not only outlook the work can be done in the next steps, but point out the potential problems and difficulties.