Synthetic Wideband Waveforms을 위한 SAR 영상 처리 기법 및 비반복성 고해상도 기법에 관한 연구

Abstract

The main aim of this dissertation is to develop a new super-resolution technique for high quality Synthetic Aperture Radar (SAR) images. Generally, super-resolution techniques can be divided into two types: hardware-based, and image-based. This dissertation develops a super-resolution technique that uses both techniques.To present a comprehensive study of methods of increasing the resolution of SAR images, I first review the fundamentals of SAR that includes a simplified generic SAR system model, a measured echoed signal in various 2-D domains, and an image reconstruction algorithm.Synthetic wideband waveforms (SWWs) were developed due to an interest in alternative technologies for achieving high range resolution. SWWs represent one such technology. Radars that use SWW do not require costly wideband hardware. Instead, high range resolution is attained by transmitting a sequence of narrowband pulses (a "burst"). However, the quality of the SAR images obtained by the conventional SAR processing algorithm was lower than expected when using the synthetic wideband signal. This was because each narrowband subpulse has a different carrier frequency term and therefore needed azimuth compression and a different range cell migration correction (RCMC). As the bandwidth (BW) of the SWW was made wider to achieve higher resolution, the reduction in image quality became more serious. The conventional range Doppler algorithm (RDA) for SWW normally did not consider the effect of this carrier frequency factor when the subpulses were synthesized. Therefore, in this thesis I propose a modified RDA procedure in an attempt to improve the quality of SAR images obtained using SWW. This proposed procedure conducts the range compression with a partial window suitable for each subpulse and then performs azimuth compression and RCMC after considering the carrier frequencies individually. Finally, the spectra of each set of compressed data are combined using the stitching method. In experiments with an automobile-based SAR system, the proposed algorithm is quite accurate in processing the synthetic wideband signals, with a resolution improvement of 20 to 30% compared to the conventional SAR processing algorithm. Moreover, if parallel processing is possible, subpulses can be processed independently and merged to obtain a high quality SAR image much more efficiently than is possible using serial processing.The nonlinear synthetic wideband waveform (NL-SWW) is a variation of SWW that suppresses grating-lobes by varying the step frequency between the pulses and by allowing overlap in the frequency of successive pulses. By concentrating the energy near the center of the frequency band, the grating-lobes as well as the range sidelobes can be controlled by conventional spectral weighting. However, generating a desirable NL-SWW can require a complicated overall system design. Spatial Variant Apodization (SVA) is a variation of the apodization technique that uses different aperture functions depending on the spatial location, and is extremely simple computationally. Super-SVA is an iterative SVA procedure

it is one of the most popular super resolution techniques due to its resolution improvement and sidelobe reduction capabilities. However, when this technique is applied to SAR images using a conventional SWW, sidelobes are controlled but the grating-lobes are not eliminated effectively. I proposed an alternate approach which reconstructs NL-SWW using super-SVA in the general SWW data. The approach involves NL-SWW implemented by reconstructing BW expanded subband spectra that is applied super-SVA. I show that the proposed algorithm has much better performance than conventional algorithms and to improve resolution slightly. Also, without recomposing the hardware, it can apply various NL-SWW methods which have been suggested previously.Spectral weighting is the easiest way to reduce the sidelobes in SAR images, but this is achieved at the expense of reducing the mainlobe width. To solve this problem apodization techniques have been suggested. Apodization significantly reduces the sidelobes while retaining a good mainlobe resolution, but has the disadvantage of an excessive computational burden. To overcome this shortcoming of the nonlinear techniques, various super-resolution techniques have been proposed. However these have high computational complexity because they entail iterative calculation. To improve image resolution without requiring iterative calculation, various inverse filtering techniques have been suggested. However, these have a serious problem in extending the BW beyond the chirp signal BW, and some methods have do not reduce sidelobes adequately, and work well only when SAR images have a sufficiently high signal-to-noise ratio (SNR). Therefore, I proposed a modified SVA method to improve SAR resolution. This method adds a modified GMF filter before SVA processing. The proposed filtering method expands the effective BW and improves the resolution while maintaining the sidelobe performance of SVA. This method has low computational complexity due to its non-iterative procedure, which differs from other super-resolution techniques. In the proposed technique, various parameters are flexibly determined based on the SNR. Thus, the proposed method can improve image resolution over a range of SNR, although this improvement decreases as SNR decreases. I tested the proposed method by simulation with point targets and in experiments with real SAR data. The proposed algorithm improved the resolution by about 40% while maintaining SVA’s ability to reduce sidelobes.