Research On New ISAR Imaging Technique

Inversed synthetic aperture radar (ISAR) can not only detect the non-corporativemoving targets at any time and locations, but also has good imaging resolution.However, it still has some faults. When the target is moving along the radar light ofsight, it cannot work. And large numbers of data assuring imaging resolution withcentimeter prevent data fast transferring and make ISAR imaging more difficult. Forthese problems, the bistatic ISAR (Bi-ISAR) and compressed sensing ISAR (CS-ISAR)imaging techniques are investigated, and the major work are summarized as follows:1. For Bi-ISAR imaging of high speed and complex motion target, a novel imagingmethod is proposed. On the range, the proposed range compression method based onfractional Fourier transform (FrFT) can overcome the blur phenomenon leaded byspectrum energy diffusing and leaking of the Fourier transform. Firstly, the rotatingangle of FrFT is estimated via the prior information of echoes and the principle ofminimum entropy of energy. Then we take the FrFT on the echoes at the estimated angle.Finally, better range compression results are obtained than the Fourier compression, anddefocus phenomenon in the range is avoided. On the azimuth, we combine the discretepolynomial phase transform (DPT) with CLEAN techniques to estimate parameters ofradar echoes. Via these parameters, the useful components of radar echoes are extracted,and the high order terms are subtracted from echoes. For those very complicatedmoving targets, the radar echoes is decomposed into some linear frequency modulatedsignal, and time-frequency analysis of these decomposed signal is accumulated. As aresult, simulated results validate Bi-ISAR imaging method for high speed and complexmoving targets.2. For the required big bandwidth and extensive sampled data of traditional ISARimaging method, a novel2-dimensional CS2-dimensional ISAR imaging method isproposed. On the range, a new sparse dictionary with phase information perserved isconstructed according to the rearranged phase expression of radar echoes. As a result, anew CS method with phase-preserved is presented. Meanwhile, the L1-Synthesismethod assures the accuracy of phase of recovered signal. On the azimuth, I use CS method to extrapolate the radar echoes beyond the observed time. Therefore, highazimuth resolution can be acquired even with short observed time. The simulated resultsvalidate this2-d CS2-d ISAR imaging method.3. To obtain the optimal results of CS problem, optimization for both sparsedictionary and the sparse coefficients of the L1norm is investigated. For the sparsedictionary, an improved Fourier sparse dictionary is proposed. Generally, the Fourierexpanding for signal of finite length will lead to the spectral leaking and widening. As aresult, the classic Fourier sparse dictionary cannot guarantee the optimum of sparsecoefficient for signal of unknown frequency in the most cases. However, the proposedsparse dictionary is obtained by the frequencies shift operation and weightedaccumulation of the Fourier sparse dictionary, which can assure the stability of sparse ofsignal. For the sparse coefficients of the L1norm, weighted optimization for sparescoefficients is taken. Finally, robust CS recovered results are obtained.