Synthetic Aperture Radar

Andrew S. Milman's Research On

Synthetic Aperture Radar

Recently, I have been developing a new method for processing SAR data, that I call omega-k Migration Processing. I published it in the International Journal of Remote Sensing, 1993, 14, pp 1965-1979. It is both more powerful and simpler than polar formatting or convolutional strip-map processing, for several reasons:

It is mathematically exact: no plane-wave approximation whatsoever involved. It takes into account the curvature of the wavefronts exactly. In essence, it does this by using a Hankel transofrm instead of a Fourier transform in the along-track direction.

Because it is mathematically exact, it provides new insights into how SAR data can be processed. One important aspect is that it leads us to a method for motion-compensating ultra-wideband SAR data.

Also because it is exact, it can be used to process an entire image at one time, rather than making sub-images and patching them together, as we do with polar formatting. Among other things, this eliminates the geometrical distortion (the keystone effect) and the phase discontinuites that arise when the sub-images are patched together.

Partly because it makes it unnecessary to patch the sub-mages together, and partly for other reasons, it is faster than polar formatting.

For strip-map SAR, it can be implemented in such a way that the data are processed only once. Methods that solve the focusing problem by creating many small sub-images require that the data be processed many times: this obviously wastes time and increases the computer requirements.

Ultra-widband SAR's can be used to penetrate foliage. However, because they must work at very low frequencies (200 to 800 MHz or so), the beamwidths are very large, about 30 degrees. In this situation, omega-k migration processing is the only efficient way to process the data.

Ultra-wideband SAR's have another problem. Normally, if the aircraft deviates from a straight path, this motion is compensated for by applying an appropriate phase shift to each pulse. But when the wave-front curvature is large, this doesn't work. Multiplying the pulse by a complex phase shifts it in a radial direction, while compensation for the aircraft motion requires a shift in the direction perpendicular to the flight track. When the curvature is large, these are quite different. There is a way to solve this problem within the framework of omega-k Migration processing.

There is increasing interest in interferometric SAR, where two complex images are compared and such things as terrain height or target motion are inferred from the difference in the complex phases of the two images. But if there are phase discontinuities in the images due to subimaging, the interfermetric information is lost at the boundaries of the subimages. This can make some otherwise easy tasks almost impossible. Omega-k Migration processing eliminates these phase discontinuities and simplifies both the generation of interferograms and their interpretation.

Unfortunately, I have no more copies of this article. It treats the geometric, but not the electromagnetic, aspects of omega-k Migration processing, which you can find elsewhere on my web site.