Description:Pattern Recognition to monitor explosive strombolian activity
Abstract:<b>Scope of the application:</b><br>
To monitor strombolian activity in order to automatically classify different types of explosions and achieve real time identification of critical ones. In this project we choose the infrared camera as image source for explosions monitoring.
<b>How it works:</b><br>
<br>The application is composed by:
<br>a Loader module which captures images to be processed (4 images per second)
<br>a Trigger module which analyzes acquired images and selects the ones suspected to be an event, basing on predetermined parameters
<br>a Scanner module which classify events founded by the Trigger, discarding wrong detections.
<br>Parallel computing is useful for multiple scanner processing in order to collect statistical data for a better understanding of volcano behavior.
<br>In this way it is possible to test a lot of different methods to achieve the best fitting for the explosion.