Analysis of thermal video for coarse to fine particle tracking in volcanic explosion plumes

Abstract

This paper presents two algorithms for feature extraction and segmentation. The first algorithm is applied to detect tens of thousands of ballistic targets moving at high velocities (100’s m/s) and with different sizes, velocities, shapes and directions. Upon detection, we compute statistics for each of these parameters for each particle, without any assumption or a-priori information. The second algorithm was developed to detect a slow moving convective cloud. The challenge was to follow the evolution of the cloud contours which comprised a heterogeneous element in front of a homogeneous, but moving (trees in wind), background. These algorithms were applied to images acquired with thermal cameras with different settings (frame rate, frame size, focal length, instantaneous field of view). A case study is presented using images of volcanic and made-man explosive events.