Research On Time Reversal Based Superresolution Electromagnetic Imaging Algorithm

The time reversal(TR) techniques based time reversal operator(TRO) super-resolution electromagnetic imaging algorithms are researched in this thesis. On one hand, as the low resolution of decomposition of the time reversal operator(DORT) method, a time reversal estimation of signal parameters via rational invariance techniques(TR-ESPRIT) method is proposed for single target fast imaging. On the other hand, considering of the low resolution of DORT method and huge computing time and memory cost of TR-MUSIC method, a fast subspace method which using signal subspace DORT method for low resolution but fast goal area pre-evaluation in the first step and using noise subspace TR-MUSIC method for super-resolution target imaging in the second step is proposed. At last, considering of the orthogonality of signal subspace vectors and noise subspace vectors, a TR-Minimum Norm method which based minimum norm ideology is proposed.Firstly, a TR-ESPRIT method is proposed for signal target fast high resolution imaging. TR-ESPRIT algorithm has the advantage of fast computing for using signal subspace and has the advantage of high resolution for employing rational invariance of ESPRIT process. Compared with signal subspace DORT method, TR-ESPRIT method has higher resolution but weaker stability under noise scenario because of the relationship between invariance and noise level.Secondly, considering of signal subspace imaging methods’ stability and strong ability under noise scenario and of noise subspace imaging methods’ high resolution advantage, a fast subspace imaging method is proposed under deep research of super-resolution model DORT and TR-MUSIC methods. In the first step, a low resolution fast goal area pre-evaluation process is completed by signal subspace DORT method. In the second step, the noise subspace TR-MUSIC method is employed for target imaging in the evaluated area under super-resolution about one-twentieth wavelength. Compared with traditional DORT and TR-MUSIC method, fast subspace method has big advantages both in computing time and memory consuming. At the same time, fast subspace method can work both in single array echo-mode and bistatic transmitting-mode.In the end, TR-Minimum norm method is proposed based on subspace orthogonality and minimum norm method. Compared with TR-MUSIC method, TR-Minimum norm method uses less noise subspace vectors and omitted the process of dividing signal and noise subspace. In strong noise scenario, TR-Minimum norm iterative method is employed to provide stable targets imaging.