A Nonparametric Approach to Segmentation of Ladar Images
Abstract
The advent of advanced laser radar (ladar) systems that record full-waveform signal data has inspired numerous inquisitions which aspire to extract additional, previously unavailable, information about the illuminated scene from the collected data. The quality of the information, however, is often related to the limitations of the ladar camera used to collect the data. This research project uses full-waveform analysis of ladar signals, and basic principles of optics, to propose a new formulation for an accepted signal model. A new waveform model taking into account backscatter reflectance is the key to overcoming specific deficiencies of the ladar camera at hand, namely the ability to discern pulse-spreading effects of elongated targets. A concert of non-parametric statistics and familiar image processing methods are used to calculate the orientation angle of the illuminated objects, and the deficiency of the hardware is circumvented. Segmentation of the various ladar images performed as part of the angle estimation, and this is shown to be a new and effective strategy for analyzing the output of the AFIT ladar camera.
Document Details
- Document Type
- Technical Report
- Publication Date
- Dec 01, 2012
- Accession Number
- ADA567488
Entities
People
- Eric A. Buschelman
Organizations
- Air Force Institute of Technology