Real-time synthetic aperture radar imaging: A model and design using VLSI correlators.

摘要

Synthetic Aperture Radar (SAR) is a two dimensional imaging system that processes reflected radar signals into an image of the reflecting object. SAR image processing is typically done using either frequency or time domain techniques. Different architectures in time domain, frequency domain and a combination of both have been studied and are summarized, compared and contrasted. However, most of the above techniques do not process the SAR data in real time to produce images.; In this research a simplified model for the time domain SAR imaging problem is presented. The model is based on the geometry of the SAR system. Using this model an expression for the phase history of the received signal is formulated. From the phase history it is shown that the range and the azimuth coordinates of a point target image can be obtained by processing the phase information during the intrapulse and interpulse period data respectively. This research also presents an algorithm for a two dimensional correlation in the time domain in order to exploit the natural parallelism to accommodate the higher computation rates necessary in time domain approaches. The algorithm requires only as many multiplier accumulators as there are non-zero elements in the correlation reference function and uses no explicit memory for its operation. It produces output at every clock cycle with a latency time of one multiply accumulate. The algorithm lends itself nicely to VLSI implementation. This research proposes a SAR architecture using these correlators to generate images in real time. A one dimensional correlator is used to process the received SAR data in the range direction and a two dimensional correlator is used to process the range correlated data in the azimuth direction. The architecture proposed has several advantages over the existing architectures such as no large memory requirements, no explicit range migration correction circuitry, efficient processing and fewer number of VLSI chips.; This research discusses the possibility of reducing the data rates by down sampling the received target azimuth phase history. It is shown that down sampling the phase history improves the spatial resolution between point targets. Down sampling and the time domain correlators help real time imaging of SAR data. The bit width requirement criterion for a given grey scale resolution at each of the processing blocks is also presented. A simulator capable of imaging point targets is also discussed.