A shift-variant restoration technique leveraging high-resolution interface modeling in lidar-based seafloor imaging
Carr, Domenic A
MetadataShow full item record
Airborne-lidar bathymetry systems provide intelligence, surveillance, and reconnaissance capabilities reliably and cost-effectively in coastal areas. Recently, systems have shown the ability to produce seafloor reflectance imagery. This imagery depends on the accuracy of seafloor coordinates and reflectance estimates comprising them. One accuracy-limiting factor is wavy sea surface topography, affecting seafloor reflectance images in two ways. First, images are geometrically distorted by uncompensated beam steering; second, images are radiometrically distorted by uncompensated pulse stretching. These two interface-induced distortions manifest themselves as image blur. This thesis presents a restoration technique addressing both of these distortions, treating them as the effects of a shift-variant linear system. This system's transfer function may be identified for each laser pulse by using a novel, hybrid-lidar processing architecture combining Geiger-mode and waveform-resolved systems' data. In this architecture, high-resolution interface models constructed from Geiger-mode system measurements are registered with co-temporal waveforms, enabling ray tracing through wavy interfaces and identifying interface-induced pulse stretching. Seafloor coordinates may then be geometrically reconstructed and reflectance estimates may be radiometrically restored. Unifying these two aspects into one restoration procedure significantly enhances the resolution of seafloor reflectance images degraded by wavy sea surface topography.