For this week, I read the papers on Moodle, met with Charlie to discuss my ideas, and read papers related to them. Below are some updates for each topic.
Idea 1 Title: Improve Fire Identification Mapping and Monitoring Algorithm (FIMMA)
Description: Enhance FIMMA to reduce the number of false-positive results and apply the algorithm to detect fire in urban areas.
Update: This algorithm has been developed for the past 20 years by NASA. It has gone through many modifications and enhancements, and is currently producing quite accurate results. The main reason why it cannot detect fire instantly is because of the time the satellites take to circle the Earth.
Idea 2 Title: Fire Detection Using A Combination of Different Image Analysis Techniques.
Description: The research aims to detect fire by analyzing different attributes (heat, temperature of the surrounding area, temperature and color of the smoke) at different ranges of the electromagnetic spectrum (UV, visible, and infrared).
Update: There have been numerous unmanned vehicles being developed to take images of a fire scene at different electromagnetic spectrum. A drawback of detecting fire using image analysis might potentially lie with the hardware rather than the analysis techniques. For example, the resolution of the images and the hardware of the camera are usually affected when the vehicles try to take photos too close to the fire source.
Idea 3 Title: Real-time Fire Tracking System.
Description: The research aims to create a system that can provide current data for active fire as well as calculate the direction it is moving. The system can combine the data from Google Earth and NASA’s Fire Map to provide the current condition of the fire. It will also need real-time data of a given location, particularly weather (wind, humidity) and local condition (population, buildings) in order to make predictions.
Update: Since last week, I have been focusing on this idea more than the other two. What I am planning to do is to build a system that can produce a simulation of how an active fire spreads in a specific fire situation, particularly in forest fire. Many factors will have to be considered to build this simulation, especially data about the surrounding environment, weather, and terrain, etc. I plan to use archived data so that I can compare my predictive result to the actual fire spread direction.