Cross-correlation registration algorithm enhancements for multipixel inverse synthetic aperture LiDAR

The accurate registration and realignment of complex signal volumes is critical for cross-range aperture gain in 3D LiDAR aperture synthesis. For targets at long range, only a limited number of diffraction-limited pixels will be projected on the target, resulting in low cross-range support. In addition, the signal-to-noise ratio (SNR) is typically low. This research describes an enhanced cross-correlation registration algorithm for 3D inverse synthetic aperture LiDAR data volumes that improves performance for low cross-range support, low SNRs, and relatively large aperture shifts. The registration performance is improved through statistical removal of the cross-correlation noise pedestal and compensation for the reduced signal overlap caused by larger shifts. The registration performance is characterized as a function of SNR, signal shift (target rotation rate), and target pixel support. The algorithm's improvements allow for registration convergence at 1-5 dB lower SNR than the baseline cross-correlation algorithm. In addition, the algorithm enhancements allow for registration convergence at 10%-20% greater shifts.