In past research, two-pass repeat-geometry synthetic aperture radar (SAR) coherent change detection (CCD) predominantly utilized the sample degree of coherence as a measure of the temporal change occurring between two complex-valued image collects. Previous coherence-based CCD approaches tend to show temporal change when there is none in areas of the image that have a low clutter-to-noise power ratio. Instead of employing the sample coherence magnitude as a change metric, in this paper, we derive a new maximum-likelihood (ML) temporal change estimate-the complex reflectance change detection (CRCD) metric to be used for SAR coherent temporal change detection. The new CRCD estimator is a surprisingly simple expression, easy to implement, and optimal in the ML sense. This new estimate produces improved results in the coherent pair collects that we have tested.

Automated synthetic aperture radar (SAR) stereo correspondence becomes increasingly difficult when imaging high-relief terrain utilizing large stereo crossing-angle geometries because high-relief SAR image features can undergo significant spatial distortions, causing a failure of traditional correlation matching. This paper presents eight coherent spotlight-mode cross-track stereo pairs with stereo crossing angles averaging 93.7deg collected over a terrain with slopes greater than 20deg. These stereo pairs suffer from terrain-induced distortions, resulting in a decrease in complex correlation (coherence) when utilizing scanning-window correlation calculations. The search to maximize complex correlation is changed from a shift-only (disparity) search to a shift-and-scale search using the downhill simplex method. This approach is tested against complex imagery with simulated distortions and then employed on the eight wide-angle stereo collects. The resulting digital terrain maps (DTMs) are compared to ground truth. Using a shift-and-scale correlation approach to estimate disparity, the relative height errors decrease, and the number of reliable DTM posts increase

Coherent cross-track synthetic aperture radar (SAR) stereo is shown to produce high-resolution three-dimensional maps of the Earth surface. This mode utilizes image pairs with common synthetic apertures but different squint angles allowing automated stereo correspondence and disparity estimation using complex correlation calculations. This paper presents two Ku-band, coherent cross-track stereo collects over rolling and rugged terrain. The first collect generates a digital elevation map (DEM) with 1-m posts over rolling terrain using complex SAR imagery with spatial resolution of 0.125 m and a stereo convergence angle of 13.8/spl deg/. The second collect produces multiple DEMs with 3-m posts over rugged terrain utilizing complex SAR imagery with spatial resolutions better than 0.5 m and stereo convergence angles greater than 40/spl deg/. The resulting DEMs are compared to ground-truth DEMs and relative height root-mean-square, linear error 90-percent confidence, and maximum height error are reported.

We introduce a general formulation for wavefront curvature correction in spotlight-mode SAR images formed using the polar-formatting algorithm (PFA). This correction is achieved through the use of an efficient, image domain space-variant filter which is applied as a post-processing step to the PFA. Wavefront curvature defocus effects occur in certain SAR collection modes that include imaging at close range, using a low center frequency, and/or imaging very large scenes. Our formulation is general in that it corrects for wavefront curvature in broadside as well as squinted collection modes, with no computational penalty for correcting squint-mode images. Algorithms such as the range migration technique (also known as seismic migration), and an enhancement known as frequency domain replication, FReD, have been developed to accommodate these wavefront curvature effects. However, they exhibit no clear computational advantage over space-variant post-filtering in conjunction with polar formatting (PF2). This paper presents the basic concepts of the formulation, and provides computer results demonstrating the capabilities of space-variant post-filtering.

A new 3D tomographic formulation of spotlight mode synthetic aperture radar (SAR) is developed. This extends the pioneering work of Munson et al. (1983), who first formally described SAR in terms of tomography but who made the simplifying assumption that the target scene was 2D. The present authors treat the more general and practical case in which the radar target reflectivities comprise a 3D function. The main goal is to demonstrate that the demodulated radar return data from a spotlight mode collection represent a certain set of samples of the 3D Fourier transform of the target reflectivity function and to do so using a tomographic paradigm instead of traditional range-Doppler analysis. They also show that the tomographic approach is useful in interpreting the reconstructed 2D SAR image corresponding to a 3D scene. Specifically, the well-known SAR phenomenon of layover is easily explained in terms of tomographic projections and is shown to be analogous to the projection effect in conventional optical imaging.