Development of a Machine Vision Fire Detection System
Abstract
This project resulted in the development, test, and delivery of a patented Machine Vision Fire Detector System (MVFDS) that provides for the first time a unique and reliable method of detecting fire events and determining their size, growth, distance, location, and overall threat in real-time. The system also provides simultaneous video coverage of the area being monitored by the MVFDS for fires. This 'man-in-the-loop' capability provides an option for manual override of automatic suppressant dump, or manual release of suppressant agent. The MVFDS is designed to be immune to false alarms based upon its decision process which involves identification, comparison, and deduction (emulates a human's process of deduction/decision) of unique properties of fire. These unique properties have been included into a fire model from which algorithms have been developed. The MVFDS uses a commercially available color CCD camera, frame grabber, microprocessor, video chip, and electronics. In aircraft hangar and facility applications, the detector is designed to identify a 2-foot x 2- foot fire at a distance of 100 feet in less than 0.5 seconds with no false alarms and, in other applications, detect fires in less than 30 milliseconds.
Document Details
- Document Type
- Technical Report
- Publication Date
- Mar 01, 1994
- Accession Number
- ADA278208
Entities
People
- A. D. Goedeke
- B. Drda
- G. Healey