Research On Lidar Signal Processing And Parameter Estimation Of Target Micro-Motion Features

Target recognition is the premise and basis for accurate combat in modern warfare.However,the current developments in camouflage,stealth,and deception have brought challenges to achieving accurate target detection and recognition.Micro-motion characteristic is one of the target motion characteristics,which reflects the inherent properties of the target.The characteristics of its uniqueness,low controllability opened up a new way for target detection and recognition,and caused the wide attention of scholars both at home and abroad.Laser micro-Doppler detection has higher sensitivity and resolution than the microwave detection.With the precise parameter estimation method and sufficient prior knowledge,the application of laser micro-Doppler effect can be extended from classification to fine recognition.In order to realize the target classification and fine recognition,based on the coherent laser detection of target micro-Doppler effect,the fields of echo model,time-frequency characteristic analysis and accurately estimation of micro-motion parameters are deeply researched through the theoretical analysis,simulation computation and experimental verification method.(1)To solve the inadequate problem of quantitative description of laser micro-Doppler signal,the accurate characterization method is proposed.The light current expression of the echo signal is derived,and the corresponding relation between the micro-motion parameters and Micro-Doppler signal parameters are analyzed.The relationship of the target's parameters,detection system parameters and the echo time-frequency characteristic are researched.The result points out that translational and micro-motion parameters are the main factors that determine the time-frequency characteristics of echoes.The Cramer-Rao bound of the parameter estimation in laser Micro-Doppler signal is derived strictly under the white gaussian noise,and the closed form expression is given,which afford an unified standard for performance comparison and evaluation of the parameter estimation methods.Through the accurate characterization of echo signals,a basic model for accurate extraction and parameter estimation of target micro-motion features is established.(2)A direct use of the existing methods on laser detected translation-micromotion feature extraction may face the problem of slow processing speed caused by large data amount.Aiming at this problem,a complete method of rapid feature extraction based on time-frequency analysis is proposed.The evaluation standard of the time-frequency analysis method is defined,which optimized the algorithm parameters of smooth pseudo Wigner-Willie distribution.The noise reduction method for time-frequency diagram is proposed based on morphological.The filter threshold is determined by making use of the one-dimensional histogram distribution,which is able to remove the noise adaptively and enhance the readability of the time-frequency characteristic.On the basis of noise reduction,we proposed the extraction method of time-frequency characteristics based on the tracing curve,which has high efficiency.The sinusoidal periodic continuation method is designed for restraining the end effect in the process of empirical mode decomposition,which has low requirement to the signal length and reduces the amount of calculation effectively.The proposed method can realize the rapid extraction of the target micro-motion features,which provides technical support for real-time classification and preliminary identification of micro-motion targets.(3)For the time-frequency overlapped problem in single-channel multi-component(SCMC)micro-Doppler signals,a parameterized multi-dimensional micro-motion parameter separate estimation method is proposed.Based on the fractional Fourier transform,the narrowest bandwidth searching method from the parameter domain is proposed,which estimated the translational parameters directly from the mixed signal.Compensating the translational component,the time-varying auto-regressive model is set up for the remaining Micro-Doppler signal.Then,the constrained particle filtering is put forward for signal separation,where the signal instantaneous frequency(IF)curves of each component are obtained directly from the time domain signal.The effect of the proposed method is not affected by the little data quantity.By solving the uncertainty caused by the underdetermined condition,the method fills the gap in the field of direct separation of TFO-SCMC micro-Doppler signals.The static parameter model particle filter method is employed to estimate the multi-dimensional micro-motion parameters from the IF curves.By designing the global cost function and invalid observation point removal mechanisms,the parameters estimation precision of our method is better than existing method for 2 ~ 3 orders of magnitude.The parametric estimation method is suitable for a micro-Doppler signal with a sinusoidal frequency modulation(SFM)model,which can separate the overlapped siganls effectively.The proposed method enriches the method for estimating the micro-motion parameters under the under-determined condition.(4)In order to estimate the micro-motion parameters directly from the echo micro-Doppler signal,reduce the intermediate process and improve the accuracy of parameter estimation,a joint parameter estimation method of laser micro-Doppler signal based on the maximum likelihood theory is proposed.The closed-form expression of maximum likelihood estimation is derived,which provides the model basis for parameter joint estimation.The singular value ratio spectrum method is used to estimate the micro-motion frequency first.The estimation precision is improved by improving the signal matrix construction rules.Then the Monte Carlo method is employed to achieve the joint maximum likelihood estimation.The simulation result shows that the estimation precision is strongly influenced by the length of the signal,and is not sensitive to the change of signal to noise ratio.In view of the disadvantages of traditional methods in calculating the laser Micro-Doppler likelihood function,a new likelihood function is designed to obtaine the ideal form of the probability density distribution.The joint estimation is implemented by Markov Monte Carlo method.The simulation and experiment show that the improved algorithm is able to achieve estimation performance approximate to the Cramer-Rao bound.The maximum likelihood-based parameter estimation method is not limited by the micro-Doppler signal model,which helps to obtain a wider application range of target recognition and a higher parameter estimation precision.The innovations of the dissertation are as follows: Firstly,the laser micro-Doppler signal is accurately characterized from three aspects,which are the echo light current model,the signal characteristic relationship and the parameter estimation accuracy;secondly,the rapid feature extraction method is proposed for laser micro-Doppler signal.Thirdly,a parameterized micro-motion parameter separate estimation method is proposed.Fourthly,a joint estimation method of the micro-motion parameters is proposed.The multi-dimension and high-accuracy fast estimation of target micro-motion parameters based on the coherent laser detected micro-Doppler effect is realized,which provides a novel processing method for the development of target recognition technology in the micro-Doppler lidar.