a pa pr m s
Feature Based Pattern Recognition - Line Finder

Typical extraction of lines using Feature-Based Pattern Recognition (FBPR)


Line segments in a square.



Lines embedded in a grassy field.



FBPR has extracted the lines.


With funding from JIEDDO, in combination with substantial IR&D support, Prometheus Scientists have used noncommutative group harmonic analysis to automatically detect and identify lines in cluttered images. This Line Finder (LF) technology has been successfully applied:

1. As an operator assistance tool for use with squint SAR images;

2. With dual-polarization SAR images and EO/IR video to discriminate vehicle tracks and treadmarks as well as power and telephone lines and their poles;

3. To imagery derived from the Vehicle And Dismount Exploitation Radar to detect vehicle treads and footprints;

4. To Yuma Proving Grounds Seeker data and foliage penetration (FOPEN) data;

5. To detect IED trip wires.

LF performs single pass detection and classification of thin wires, which may be curved and/or partially occluded, and significantly outperforms algorithms based on Radon-Hough or edge detection. The underlying mathematics is presented in the monograph Group Filters and Image Processing (Psypher Press, 2003) written by two Prometheus Principal Scientists, Myoung An and Richard Tolimieri."




© Prometheus Inc. All rights reserved. Web Privacy Statement