Multiscale Local Map-Drift-Driven Multilateration SAR Autofocus Using Fast Polar Format Image Synthesis

Abstract

This new autofocus method is based on multilateration by ranging to small target areas at independent directions on the ground. Range-clipped Doppler low-filtered profiles around target points are used to compute local images using frequencydomain polar format algorithm. Images obtained from adjacent subapertures are registered, and the displacements yield elevation, trajectory, or clock drift (bistatic case) errors. To alleviate the insensitivity of map drift to error fluctuation faster than subaperture duration, the algorithm is reiterated with coarse-to-fine resolution, yielding high to low frequency errors. This allowed true bistatic synthetic aperture radar (SAR) autofocus (without monostatic image), autofocus in circular SAR on remote areas, and, as a side product, our first successful air-to-air inverse SAR high-resolution imaging.Index Terms-Airborne radar, bistatic synthetic aperture radar (SAR), focusing. I. CONTEXTL OW TO MEDIUM (1 m to 30 cm) resolution synthetic aperture radar (SAR) imaging at Office National d'Études et Recherches Aérospatiales (ONERA) in the 80s and 90s (before high-accuracy GPS-hybridized inertial measurement units allowed direct SAR synthesis from measured trajectory) used local map-drift [1] autofocus. This technique, measuring distortion field between two (or more) independent looks (stripmap images), produced estimates in low frequency trajectory errors (typical inertial unit fluctuations were longer in time than the typical integration time needed for these resolutions), and images were synthesized again with corrected trajectory [2].High-resolution (10 cm) SAR imaging initiated in 2000 at ONERA requires autofocus for estimating velocity error fluctuations much shorter than the integration time; thus, an autofocus based on multilateration from phase tracking of pointlike reflectors was designed [3] with automatic detection of the appropriate stable pointlike reflections on the ground. This algorithm uses range-clipped and narrow Doppler-band Manuscript filtered profiles around the pointlike reflection to measure and unwrap the echo phase. From these phase variation measurements in at least three independent directions for any point of the acquisition trajectory, an accurate trajectory update can be derived.With circular SAR imaging with high banking angle and bistatic SAR imaging (and some wild remote areas without enough appropriate man-made opportunistic reflectors), this autofocus method failed. The reason for circular SAR is that high banking angle masks GPS satellites, thus ruining GPS hybridizing, raising the Doppler jitter of the point echoes which implied widening the Doppler filter, and hence possibly mixing the tracked point echo with neighboring echoes. For bistatic SAR, very few (if any) pointlike echoes are usually present due to the fact that 90 • corners (present in many man-made landscape features) scatter the radio waves back to the transmitter and not to the sensor when there is any significant bistatic angle. Furthermore, true bistatic autofocus (not relyin...

Autofocusing SAR images via local estimates of flight trajectory

“…The whole data frame after RC and RCMC is divided on short data blocks in the azimuth direction. Each data block is split into two halves and two SAR images for the MDA estimation are formed using the dechirp-based processing (11) and (12). From the image shifts (15), the local quadratic phase errors and the corresponding Doppler rate errors (14) are evaluated for each azimuth block for a set of range blocks.…”

“…The multiple-aperture MDA [3,10] estimates the phase error as a higher-order polynomial. Other models, such as a Fourier series, have also been proposed [10][11][12]. The problem is that the parametric autofocus methods are restricted by the models applied.…”

Factorized Geometrical Autofocus for Synthetic Aperture Radar Processing

“…It should also be stressed that multilateration techniques based on prominent point phase tracking [4] and local 2-D MD [5] do update the aperture track, providing an ability to correct residual space variant RCM. This has, however, not been demonstrated in practice.…”

A Robust Motion Error Estimation Method Based on Raw Data

“…These update techniques are mainly based on MDA [4], [14], [16] or PGA [5], [6]. In [4], the author divides subapertures in azimuth and estimate the line-of-sight (LOS) displacement of plane by a modified MDA in each subaperture.…”

“…In [16], the residual trajectory errors are expanded into a Fourier-like sum of one linear and a few sinusoidal error terms, and then the reduced set of error parameters is estimated by computing the registration of small rectangular patches. In [14], the multilateration technique is used to update the trajectory parameters. A linear equation set is used to calculate the velocity corrections by the apparent displacement.…”