SEVENTH
ANNUAL 

Minimum Entropy SAR Autofocus Ali F. Yegulalp MIT Lincoln Laboratory 244 Wood Street Lexington, MA 02420 tel: (781) 9810886 email: yegulalp@ll.mit.edu Abstract A common problem for synthetic aperture radar (SAR) is focus degradation due to unknown or uncompensated platform motion. Algorithms that adaptively correct such focusing errors are generically referred to as autofocus algorithms. In this talk, we present a new autofocus algorithm based on the minimumentropy principle. Unlike other autofocus algorithms, the minimumentropy approach does not rely on using pointscatterers in the image to estimate phase errors. Instead, it searches a parameterized space of phase corrections using entropy as a measure of focus quality. Consequently, very little image contrast is required to focus images. In fact, substantial phase errors can be reliably corrected even when the image consists of pure noise (provided the noise is not exactly Gaussian). With reasonable contrast levels in the image, extremely severe phase errors can be routinely corrected. We will show examples of minimumentropy autofocus applied to both real and synthetic
SAR images. We also compare minimumentropy to phasegradient autofocus (PGA), currently a
popular choice for SAR autofocus. PGA relies on finding point scattering centers in the
image, so it performs poorly with severely defocused or low contrast images. PGA is also
incapable of correcting twodimensional focusing errors, which is easily done using
minimumentropy. On the other hand, PGA requires less computation than minimumentropy. 
Direct comments and questions to: webmaster@ll.mit.edu
© MIT Lincoln Laboratory. All rights reserved.