Detecting unknown chemical clouds at distance with multispectral imagery

Past chemical warfare agent attacks often mention the use of mixtures of chemicals or impure-incomplete formulation. Terrorist groups may also generate new chemical toxic agents. Those situations involve unknown compounds and thus may be undetectable by traditional methods. Indeed, standoff gas detection with infrared devices traditionally relies on the comparison between measured signal with a library of signals included in a database. Observing the gas absorption in infrared band III (LWIR 8-14 μm), our multispectral infrared camera is used to detect gas clouds up to a range of several kilometers, to provide identification of gas type and to follow the motion of the cloud in real time. The approach described in this paper develops an algorithm that enables the device to detect gas even if the measured signature is not in the database – pattern-matching-free algorithm. This detection process has been evaluated in the laboratory and subjected to significant experimental feedbacks. The results are a capability to detect unknown gases and gas mixtures.