| Synthetic aperture radar?SAR?is of major importance in the field of remote sensing.Due to the joint efforts of these scholars,since the first successful operation of a synthetic aperture radar in 1950s,synthetic aperture radar system design,operational performance,processing algorithms have been improved.Its applications are continuously expanding,and gradually used for more complex scenarios.However,the newly-developed applications do challenge the existing techniques and algorithms.First,the operation of spaceborne and airborne SAR system costs great.Therefore,based on the optimization of the conventional technology,the development of high-precision and high-efficiency SAR raw data generation technology is very necessary.Second,the SAR imag-ing algorithm customed for newly-developed optronic processor may fail when processing the satellite-based low-squint?Squint angle|?|<?H/2,?His beamwidth?[1]SAR data.Therefore,based on traditional SAR imaging techniques,the development of SAR imaging algorithm cus-tomized for programmable optronic processors to process both broadside and low-squint SAR data is very necessary.Then,the spectrum may sometimes wrap around azimuth frequency as a result of ambiguity.Therefore,new feature extraction technique should start with raw data and free from the Doppler spectrum,the development of non-ambiguity signal-level fea-ture extraction technique is very necessary.Finally,moving target extraction from dual-channel SAR image is a very important SAR application and better methods are expected in this area of researching.Therefore,the development of new dual-channel SAR moving target extraction algorithm is very necessary.In this paper,the presented technical challenges have been carefully analyzed,our inno-vative work is mainly reflected in optimizing traditional technology and developing a series of new information extraction technology facing those complex scenarios.The main contributions of this paper can be summarized as follows:First,we optimize the traditional raw data generation technology,and propose a novel raw data generation technology based on GPU acceleration and RCS calculation.The technology relies on GPU parallel computing and echo symmetric matrix optimization algorithm,which greatly enhances the SAR raw data generation technology in operational efficiency.Meanwhile,the introduction of a high-frequency approximation in RCS calculation enables us to obtain the backscattering coefficient based on the instantaneous slant range and incident angle.Given Dig-ital Elevation information?DEM?and radar incidence angle range,the variant backscattering coefficient of each scattering centers within the synthetic aperture time can be accurately calcu-lated.Thus,relative to traditional SAR raw data generation technology,this technique further enhances the accuracy of generated raw data.Second,the traditional optronic SAR imaging algorithm has been improved,and we present a novel SAR imaging algorithm customized for programmable optronic processor in the ap-plication of general SAR image formation for broadside and low squint angle?Squint angle|?|<?H/2,?His beamwidth?.In the optronic implementation,a large volume of SAR data is focused block by block and then their imaging outputs are mosaicked into a SAR image.By applying signal-domain preprocessing filter,signal spectrum is shifted back to zero azimuth fre-quency.It allows the SLM-loading phase filters to cover the most part of signal spectrum,and thus guarantees a high-quality imaging result.The frequency-domain phase filtering enables the imaging result to be centrally located in light output.Thus,the subsequent image mosaic can be accomplished with minimum system cost and power consumption.This algorithm also ensures the imaging result of broadside SAR data to be well-focused.Therefore,the optical approach can guarantee a high-quality SAR image output when a large volume of SAR data in either broadside case or low-squint case is processed.Third,through deep research on signal-level information extraction,we propose an inno-vative algorithm based on M-RANSAC and STFRFT to realize feature extraction.To avoid the spectrum wrapping,this algorithm starts with time-domain range compressed signal.Therefore,this algorithm establishes its principal contribution as the ambiguity-free signal-level feature ex-traction of scattering centers.The extracted feature includes:?1?the backscattering envelope which can be the feature of major concern to characterize target properties;?2?the geograph-ical location which denotes the cross-track and along-track positions;?3?the relative velocity between scattering center and radar platform which reflects the along-track speed.It is worth noting that the realization of feature extraction without knowing explicit knowledge of platform velocity and forming a SAR image provides additional novelty of this algorithm.Fourth,after deep research on image-level moving target extraction,we report a novel two-step dual-channel GMTI algorithm in this paper.First,jointly considering the strength and velocity of movers,a new metric,called weighted velocity,is presented.The corresponding weighted velocity locator?WVL?is capable of outputting detecting results without the noise-related false alarms.Then,a postprocessing filter,named false alarm canceller?FAC?,is im-plemented to eliminate the remaining clutter-related false alarms.Thus,a fine detecting result can be achieved by this two-step GMTI method.In this article,we also study the pre-processing technology for image-level information extraction,and present a multi-channel SAR imaging and registration technique based on signal modeling.This technology lies a solid foundation on above mentioned technology:image-level moving target extraction. |
PDF Full Text Request