A Novel Highly Parallel Algorithm For The Detection And Tracking Of Chemical Gas Plumes Using Hyperspectral Video Sequences

Multitemporal hyperspectral images provide detailed spectral information that makes them particularly suitable for the detection and tracking of chemical gas plumes. However, due to the high dimensionality of the data to process, classical video processing techniques do not ensure a real-time performance, which makes imperative the development of new advanced and computationally efficient algorithms. In this paper, we propose a novel method for the detection and tracking of chemical gas plumes, recorded in hyperspectral video sequences, which has been specially designed for being easily parallelizable in hardware devices for applications under realtime constraints. The identification of the gas plume presence has been tackled as a spectral change detection problem using the temporal redundancy of two consecutive frames. To maximize spectral differences, the first derivative of each frame pixel is computed and compared with its homologous pixel in the next frame measuring the spectral angle between them. We have applied the proposed methodology to two real video sequences. The results obtained support the benefits of our proposal.