Iterative Signature Suppression for detection and identification of gas plume constituents

In the case of long-wave infrared (LWIR) hyperspectral images, the detection and identification of all gas constituents in a plume is not well achieved by common detectors such as the Adaptive Cosine Estimator (ACE) and the structural background Generalized Likelihood Ratio Test (GLRT). This paper proposes a new approach for hyperspectral imaging called “Iterative Signature Suppression (ISS)” based on the iterative use of the structural background GLRT to detect all gaseous mixture components. The idea behind this method, is to add a previously identified gas signature to the background sub-space in order to“subtract” this signal by applying the GLRT again. The results have shown that the ISS method can improve the detection of gas plume constituents hidden under the strongest gas in the plume.