Hawa Yves MEEOH27, pp. 62. TEK/avd. för signalbehandling, 2008.
In this thesis, we present a study of the Fast Backprojection (FBP) known to be a fast time-domain algorithm for image retrieval in Synthetic Aperture Radar (SAR). As a time domain algorithm, FBP possesses inherent advantages such as perfect motion compensation, unlimited scene size, wide bandwidth and ability to handle long integration angles. Although FBP reproduces SAR images on pixel-by-pixel basis, the processing time for FBP is reduced significant compared to the Global Backprojection (GBP) with a heavy computational load. For GBP, the number of operations to process a N×N SAR image with N aperture positions is proportional to N . Whereas, the number of operation required by FBP in the same case is reduced by a factor of sqrt(N) i.e. proportional to N2sqrt(N). We give a detailed explanation on how to implement FBP in Matlab in order to retrieve a SAR image. We explain about different techniques of interpolation used to obtain a high quality image. We also present a method to compare between interpolation techniques in term of SAR image quality and processing time. A comparison between FBP and GBP is also a topic to be discussed in this thesis. In this comparison, the processing time in theory and reality will be focused.