Hughes Aircraft Company

Robert T Frankot

Ralph E Hudson

Robert H Latter

George H Senge

Ramesh Rampertab

Gene W Zeoli

A method of providing accurate elevation data from three or more complex SAR images. A SAR vehicle is operated to produce a number of SAR images at different grazing angles. The present method provides precise elevation maps derived from the SAR images. In a multiple channel variant of the present invention, more than two antennas collect SAR data from various grazing angles, increasing sensitivity while resolving the ambiguity problem in a mathematically optimum manner. A multiple pass variant of the present invention collects data from more than two passes, and from various grazing angles (interferometric 'baselines'). The multiple pass approach and associated processing preserves the high sensitivity of the long-baseline available from dual pass IFSAR while resolving the ambiguity problem in a mathematically optimum manner. Accuracy and ambiguity resolution are improved with each additional pass or channel. The method processes raw SAR data to produce a plurality of complex images therefrom. The complex images are processed to produce a plurality of linear phase matched complex images and predetermined processing parameters. The plurality of linear phase matched complex images are then processed to produce a elevation map. The plurality of linear phase matched complex images are produced by matching noise power contained in each of the complex images to normalize the images to compensate for differences in gain of the systems from scan to scan, matching linear phase values of each of the complex images to correct for phase differences by estimating phase matching parameters for each of the complex images, estimating a phase-difference scale factor for each of the complex images, and ordering each of the complex images in terms of increasing phase difference scale factors. The linear phase values of each of the complex images are matched by computing a pairwise interferogram for adjacent complex images, summing sample-pair products of the interferogram to estimate the linear phase thereof, and computing the phase of the summed sample-pair products. The elevation map is computed by computing an initial elevation map using conventional IFSAR techniques applied to a pair of complex images, estimating additive phase correction values for each of the linear phase matched complex images, and estimating elevation values by maximizing the magnitude of the complex weighted sum of M complex images or complex interferograms, where the weights include the phase attributable to the estimated elevation to produce the elevation map.

Multi-pass and multi-channel interferometric synthetic aperture radars

A method for automatically selecting subareas from reference image data such that registration accuracy is optimized between images. Optimal subarea selection reduces the on-line computation and reference data storage required for multi-subarea correlation or, alteratively, improves its effectiveness. Example results for a synthetic aperture radar (SAR) image are described. The results indicate that the automatic subarea selection method of the present invention reduces on-line computation by a factor of 2 to 3 (relative to random subarea selection) without degradation in accuracy. The present selection method minimizes the predicted total mean squared registration error (MSE). The total MSE is predicted in terms of the position and predicted measurement covariance (derived from local image statistics) of each candidate subarea. Combinatorial optimization procedures select a predetermined number of subareas to minimize total MSE. The present method jointly considers the quality and position for each local measurement, optimally handles any order polynomial model, and includes bounds on the uncertainty of registration parameters.

Automatic subarea selection for image registration

Apparatus including a multi-scale adaptive filter for smoothing interferometric SAR (IFSAR) data in areas of low signal-to-noise ratio (SNR) and/or coherence while preserving resolution in areas of high SNR/coherence. The multi-scale adaptive filter uses simple combinations of multiple linear filters applied to a complex interferogram. The multi-scale adaptive filter is computationally efficient and lends itself to parallel implementation. A pyramid architecture comprising a plurality of cascaded stages is employed which reduces the computational load and memory required for implementation of the processing algorithm. The multi-scale adaptive filter implements a processing algorithm that may be applied to standard IFSAR data. Its input is a complex interferogram (the conjugate product of two complex images) and its output is a filtered interferogram (A) which is passed to an information extraction processor, that extracts a terrain elevation map, for example. The adaptive filter incorporates linear filters at two or more scales (i.e. filter impulse response widths) whose outputs are combined in a data-dependent manner. The combination rules result in an output interferogram (A) that is filtered heavily in areas of low coherence and receives little or no filtering in areas of high coherence. The combination rules use a coherence measure that is a simple nonlinear function of the linear filter outputs themselves.

Multi-scale adaptive filter for interferometric SAR data

The present invention is an interferometric SAR system and processing method that combines multi-pass SAR interferometry with dual-antenna SAR interferometry to obtain elevation maps with accuracy unobtainable by either method alone. A single pass of the dual-antenna system provides coarse elevation maps. High accuracy maps are obtained through additional passes, with accuracy determined by the number of passes. The processing method combines the acquired data to provide a calibrated, high precision, low ambiguity elevation map, using approximate least-squares and maximum-likelihood processing methods. The present dual-antenna SAR interferometer collects two complex SAR images from slightly different elevation angles on a single pass using two antennas on the same platform. The present invention provides calibrated maps that have coarse precision but are nearly unambiguous because of the small interferometer baseline. The multi-pass method collects two or more complex images using multiple passes of a radar platform with each antenna. Alone, the multipass method provides much more precise, but ambiguous and uncalibrated, elevation maps. However, the present invention combines the dual-antenna and multi-pass techniques to provide unambiguous and highly precise maps.

Hyper-precision SAR interferometry using a dual-antenna multi-pass SAR

A terrain height radar system and processing method comprising a high resolution synthetic aperture radar (SAR) mounted on an air vehicle and a SAR signal processor containing a signal processing algorithm or method for computing terrain height and radar backscatter power. The system contains motion sensing and navigation functions that also provide data to the signal processor to provide motion compensation. Signal processing algorithms in the method compensate for planar motion of the air vehicle for variations of terrain height in the field of view. The algorithms also compensate for nonplanar motion of the radar, and for scatterers in or very near to a reference plane in the field of view. The algorithms exploit defocusing due to displacement from the reference plane to estimate the terrain height above the reference plane. The algorithm is computationally efficient because the bulk of the radar signal processing is common to both the SAR function and the terrain height estimation function. The following processing method is implemented. The synthetic array radar is operated to produce out-of-plane motion and generating radar return signals derived therefrom. The radar return signals are processed to compensate for relative motion, generate ground plane polar format data, generate range compressed data, and generate autofocused data. Azimuth compressed data is generated by focusing uncompressed azimuth data for each of many discrete terrain height reference levels, compressing the focused data using a weighted FFT, temporarily recording the magnitude of each pixel, storing separate image magnitude values corresponding to each terrain height reference level for each pixel and outputting the largest of these magnitudes for each pixel as a radar cross section value of that pixel, and outputting the height reference level at which that maximum occurred as the height of that pixel. Interpolation may be used to refine the height estimate. The strong scatterer signal is then canceled by estimating the amplitude phase height and azimuth of the strong scatterer, using these parameters to determine a contribution in the uncompressed data set due to the strong scatterer, subtracting this contribution from the uncompressed data, and repeating the amplitude phase height and azimuth height estimating step to provide for for improved accuracy.

Growing community of inventors

Van Nuys, CA, United States of America

Robert T Frankot