Elettronica ed elementi hardware impiegati nel sistema intelligente multi-sensore per la rilevazione e il monitoraggio della cenere vulcanica attraverso il processamento delle immagini

Abstract

Within the framework of Ash-RESILIENCE, we are performing an innovative and extremely low cost, low energy consumption and small size wireless electronic multi sensor system for the detection and monitoring of volcanic ash in sensitive areas, in order to manage information and alerts caused by ash fallout in real time. We want to implement a Wireless Systems Network in order to record, process and share data acquired to monitor areas affected by ash fallout. Collected data will be available through remote access connecting the systems directly to the acquisition center at INGV-OE. To obtain an embedded device with appropriate computational power without influencing system costs and dimensions, we employed a “Single Board Computer”, such as Raspberry Pi. The main challenges are:
• decreasing the overall complexity of the system using fewer sensors and mechanics to maximize reliability; • promoting innovative software solutions, using powerful open source image processing tools. We developed an algorithm for ash detection to receive alerts and monitoring data when the ash amount in the ground exceeds a pre-specified value of interest. From this point of view, to detect volcanic ash we used an image processing approach, which represents the central aspect of the system. To distinguish images pixels into two classes data, background (white collecting surface) and foreground (dark ash particles), we determine an automatic adaptive image threshold to carry out the binarization of images to get dependable information for the measure of the ash quantity starting with the ash particles pictures acquired by a visible, small size, low cost Full HD camera. The next goal will be integrating into the algorithm the innovative method “Pixel Digital Weight” we conceived, to accurately measure the ash weight, which does not require sophisticated scales and sensors, but only by means of the image processing.