Automatically assigned DDC number: 6298932
Manually assigned DDC number: 00637

Number of references: 4

Title: Visual Ordnance Recognition System Fiscal 1998 Year-End Report: Algorithms and Hardware Design

Author:

Subject: Clark F. Olson,Todd E. Litwin,Roberto Manduchi,Larry H. Matthies Visual Ordnance Recognition System Fiscal 1998 Year-End Report: Algorithms and Hardware Design

Description: This report describes the progress that has been made during this period, as well as our plans for 1999. We have concentrated on detection of BLU-97 ordnance, which is in current usage in U.S. military test ranges (Fig. 1). The body of this type of ordnance is cylindrical and it is 20 centimeters long, with a 6 centimeter diameter. When new, the ordnance is bright yellow in color, but it is often weathered in practice on the test range. Figure 1: Image of BLU-97 ordnance acquired at a Nellis Air Force Base test range . The scenario for which this system has been designed is as follows. A unmanned vehicle traverses a test range at a speed of approximately 5 mph following a pre-determined route in order to fully cover the test range. As the vehicle is traversing, the ordnance recognition system continually examines the terrain in front of the vehicle looking for instances of BLU97 ordnance. The ordnance recognition system provides a video signal of the terrain in front of the vehicle, visually marking the locations of the detected ordnance. When ordnance is detected, the vehicle will stop (and retreat to a safe stand-off distance, if necessary), in order to perform remediation. We have considered a large suite of algorithms to perform the ordnance recognition in this scenario, which have been divided into four categories: pre-processing, hypothesis detection, verification modules, and evidential reasoning. Figure 2 shows the flow of information between algorithms in these categories, as well as the algorithms in each category that have been examined. The images are first pre-processed in order to yield better detection performance. Candidate ordnance locations are then identified using one or more hypothesis detection methods. Verification modules are applied to the candi...

Contributor: The Pennsylvania State University CiteSeer Archives

Publisher: unknown

Date: 1998-12-18

Pubyear: unknown

Format: ps

Identifier: http://citeseer.ist.psu.edu/142336.html

Source: http://robotics.jpl.nasa.gov/people/olson/papers/jpl-d-16331.ps.gz

Language: en

Relation:

Rights: unrestricted

<?xml version="1.0" encoding="UTF-8"?>

<references_metadata>

</rec>

</rec>

</references_metadata>

Automatically assigned DDC number: 6298932Manually assigned DDC number: 00637

Automatically assigned DDC number: 6298932
Manually assigned DDC number: 00637