FLIR/LADAR Fusion for Target Identification.
Abstract
Rockwell International's objective was to develop a robust and state of the art FLIR/LADAR target detection and identification system for the reconnaissance, surveillance, and target acquisition program. The algorithm suite was to be integrated into the Unmanned Ground Vehicle (UGV) platform. But due to program changes such as the late availability of the LADAR sensor unit and funding restrictions in calendar year 1996, the primary goal was able to be addressed in this contract. This report document the major successes and conclusions that were obtained in the process. Rockwell's major successes during the contract period are: Developed a new FLIR/LADAR ATD/I Framework: A foundation for an innovative ATD/R/I system was developed for others to use. The new approach incorporates state of the art techniques such as FLIR/LADAR feature level fusion with clutter suppression and hierarchical classification algorithms. Developed FLIR-Based Background Suppression Software. It incorporated many of the FLIR-based background suppression ideas in this report. During the process of planning, research, and the development of algorithms towards the objective, the following conclusions were reached: To obtain high detection, low false-alarm rates, and robust identification of targets, one must deal with background suppression at the onset. It must be integrated into any planned ATD/R/I system. Ground rules must be established on the difference between targets and clutter objects. Humans recognize objects first as categorical levels. Hierarchical classification techniques that were developed under this contract, learn categorically. The approach shows much promise in advancing ATD/I technology. Fusing FLIR and LADAR data into a common feature vector as discussed in this report is a powerful method in exploiting the input data.
Document Details
- Document Type
- Technical Report
- Publication Date
- Sep 01, 1996
- Accession Number
- ADA316077
Entities
People
- Thomas Jenks