A Novel 2-Stage Approach For Co 2 Gas Detection Using Hyperspectral Imagery

Efforts to regulate carbon emission from human activities have increased significantly in recent years. Authorities seek remote sensing based solutions that would allow them remote monitoring of carbon emission sources. Hyperspectral imaging has emerged as an alternative technology to the classical sensor based gas identification methods. Earlier hyperspectral imaging based studies typically utilized the entire gas spectrum in their analysis for gas detection task. In this study, we propose a 2-stage approach to CO 2 gas detection in hyperspectral imagery. First, we generate a rough CO 2 detection map using only a narrow region of the CO 2 signature. Next, using a more refined CO 2 signature obtained from the image based on the high confidence regions of the first stage, we generate a refined CO 2 gas detection result. Using a mid-wave infrared (MWIR) hyperspectral camera images, we have shown the effectiveness of the proposed method in the CO 2 gas detection task.