Research in Computer Vision for Autonomous Systems
Abstract
This report addresses FLIR processing, LADAR processing and electronic terrain board modeling. In our discussion on FLIR processing, we have analyzed the issues of classifiability of FLIR features, computationally efficient algorithms for target segmentation, metrics, etc. The discussion on LADAR includes a comparison of a number of different approaches to the segmentation of target surfaces from range images, extraction of silhouettes at different ranges, and reasoning strategies for the recognition of targets and estimation of their aspects. Regarding electronic terrain board modeling, we have shown how the readily available wire-frame data for strategic targets can be converted into volumetric models utilizing the concepts of constructive solid geometry; we then show how from the resulting volumetric models it is possible to generate synthetic range images that are very similar to real LADAR images. We also show how sensor noise can be added to these synthetic images to make them even more realistic. Keywords: Vision; Pattern recognition; Audio visual system; Robotics; Artificial intelligence.
Document Details
- Document Type
- Technical Report
- Publication Date
- Sep 15, 1988
- Accession Number
- ADA212420
Entities
People
- Avi Kak
- Keith Andress
- Mark Yoder
- Steve Blask
- Tom Underwood
Organizations
- Purdue University