| Inherent in the process of forming a synthetic aperture radar (SAR) image of terrain is the presence of fading noise, which appears as speckle in the resulting picture. The quantification of the effects of fading noise has been accomplished herein for two specific applications: stereo radar height estimation and map matching guidance. While for height estimation fading has a strong, negative influence on accuracy and precision, fading does not render a SAR image useless for map matching. If the method of image comparison used is assumed to be cross-correlation, then two interesting results are obtained. First, the cross-correlation function R(,xy)((tau)) as defined by an expected value (ensemble average) is seen to predict the same peak value, no matter whether the images are noiseless, N = 1, N = 4, etc., that are being cross-correlated, due to the assumption of a certain (chi)(,2N)('2) multiplicative fading model and signal and noise statistical independence. When cross-correlation peaks are estimated using FFT techniques, it is observed that the fading noise variance affects the variance of the cross-correlation peak. Fading noise effects are compared (as to their deleterious action on the cross-correlation process) to the effects of scene rotations and to scene changes caused by backscatter changes for diffuse scattering regions. |
