Study On Obstacle Detecting Key Technology Of InSAR Based Unmanned Ground Vechicle

Synthetic aperture radar interferometry(InSAR)technique has been employed to construct digital elevation models(DEMs)of terrain surface with high spatial resolution and accuracy.As the SAR systems provide their own illumination,they can acquire information almost independently of meteorological conditions and sun illumination.This technique has been widely applied in the field of surveying and mapping.Off-road scene awareness is a challenge problem to be solved by unmanned ground vehicle(UGV).Traditional sensors,such as: optical camera,lidar and infrared detectors,have limited performance in off-road environments,especially when dealing with rough surfaces and obstacles.In this paper,we focus on perception ability of novel forward imaging interferometric synthetic aperture radar(InSAR),which can provide three-dimensional scene information of unstructured environments for UGV.In this dissertation,the key imaging technologies for the InSAR systems are studied based on the platform feature of the UGV,and the major work is focused on multi-baseline phase unwrapping(PU)and the special image feature of obstacles.The main content of this dissertation is summarized as follows:1.The multi-baseline PU method based on chinese remainder theorem(CRT)is studied.As we know,the classical CRT reconstructs an integer from its multiple remainders that is well-known not robust in the sense that a small error in a remainder may cause a large error in the reconstruction.And theoretic analysis shows that the baseline parameter and the residue make the PU method based on classical CRT not robust enough and inapplicable in practical cases.To sovle this probem,the greatest common divisor has been carried out to build the new system of simultaneous congruences,and two robust CRTs(the searching based robust CRT and the closed-form robust CRT)could be applied to multibaseline PU for the new system of simultaneous congruences,the principle of two multibaseline PU method is studied.Finally,we compare the performances of the classical CRT,the searching based robust CRT and the closed-form robust CRT algorithms in terms of both theoretical analysis and numerical simulations.The results demonstrate that the searching based robust CRT and the closed-form robust CRT has the better performance than the classical CRT.2.The searched-form cluster-anlysis based multi-baseline PU method is studied.In conventional searched-form cluster-anlysis based multi-baseline PU method,all pixels are clustered into different groups according to their recognizable mathematical patterns,and then,information of the cluster center is used to unwrap the phases of pixels group by group,which is different from pixel-by-pixel methods.Since the conventional searched-form cluster-anlysis based multi-baseline PU method use the histogram method to cluster pixels into group,it is unable to identify clusters whose intercepts are very close to other clusters when phase noise is present.In that case,a linear combination method is adopted to decrease the number of resulting groups and widen the intercept distances between each group in intercept.The simulated experiments demonstrate that the refined method is not only effective and efficient but also more robust to noise than the conventional method.3.The closed-form cluster-anlysis based multi-baseline PU method is studied.The phenomenon of the pixels belongs to the ghost clusters is existed,and the searched-form cluster-anlysis based multi-baseline PU method can't eliminate the ghost cluster centers.In this dissertation,a closed-form cluster-anlysis based multi-baseline PU method is proposed,which is based on the ambiguity d-value set and the intercept envelopes of the resulting histogram.In our method,the real cluster can be known when all ambiguity vectors are obtained.And the ghost clustesr can be eliminated because the clusters are known in advance.The phenomenon of those pixels belong to the ghost clusters is disappeared after taking a special filtering measure.The closed-form cluster-anlysis based dual-baseline phase unwrapping described earlier is extended to multibaseline cases.Results of the experiment are shown to verify the effectiveness,efficiency,and noise robustness of the closed-form cluster-anlysis based multibaseline phase-unwrapping method.4.The image feature of obstacles acquired by the InSAR systems is studied.The forward imaging InSAR is mainly difference in the platform to traditional ones,so the basic signal processing flow is not much different from that of an airborne InSAR.An InSAR signal processing flow is tailored to acquiring the height information of the scene for the UGV.Simulation and real data experiments have been carried out on terrain scene where there are obstacles to validate the proposed signal processing flow and the perception performance of the InSAR,Based on the above experimental results,the scattering feature,coherent feature and elevation feature of negative obstacle and positive obstacle are analyzed,which lay a strong foundation for negative obstacle and positive obstacle feature extraction and multi-feature detection.