Processing of FLIR Data on Dicifer.
Abstract
This effort has investigated the extent to which automatic tactical target cueing on selected FLIR (forward-looking infrared) imagery can be accomplished using digital image processing and automatic pattern recognition techniques. The DICIFER (Digital Interactive Complex for Image Feature Extraction and Recognition) system was utilized to analyze FLIR image data for the U.S. Army Night Vision Laboratory (NVL). DICIFER is a general purpose R and D tool developed by Pattern Analysis and Recognition (PAR) Corporation for Rome Air Development Center (RADC) Intelligence and Reconnaissance Division (IR). A digital process for automatic detection of tactical targets in FLIR images was designed and tested. The total process included noise removal and contrast enhancement preprocessing; gradient edge detection and boundary chain encoding; and the design of Boolean classification logic. The classification logic was applied against a test set of 34 images. The probability of detection achieved was 88% (38 of 43 targets detected). (Author)
Document Details
- Document Type
- Technical Report
- Publication Date
- Jul 01, 1976
- Accession Number
- ADA037136
Entities
People
- Dennis J. Lucey
- George E. Forsen
- Michael J. Zoracki