A SAR imaging technique for the target of complex azimuth envelope based on information extraction

This paper talks about an innovative method for constructing SAR image from raw data when range migration phenomenon is obvious. Here, SAR raw data works as input. A set of range migration curves (RMC) will be extracted from its range compression result. After that, parameters gained from curve fitting provide physical location and normalized backscattering coefficient of targets, and thus a SAR image is constructed. In this algorithm, Random Sample Consensus (RANSAC) offers idea about pixels classification of different point targets. Moreover, the least-squares distance gives measure to classify pixels belonging to different point targets. It is shown that range migration curve can be regarded as a parabola when estimating the parameters by curve fitting. These estimated parameters construct SAR image well when the conventional SAR image formation algorithm fails in obtaining a good one. In this paper, experimental results are conducted to illustrate the superiority of the algorithm.