Extremely Robust Remote-Target Detection Based on Carbon Dioxide-Double Spikes in Midwave Spectral Imaging

Infrared ship-target detection for sea surveillance from the coast is very challenging because of strong background clutter, such as cloud and sea glint. Conventional approaches utilize either spatial or temporal information to reduce false positives. This paper proposes a completely different approach, called carbon dioxide-double spike (CO₂-DS) detection in midwave spectral imaging. The proposed CO₂-DS is based on the spectral feature where a hot CO₂ emission band is broader than that which is absorbed by normal atmospheric CO₂, which generates CO₂-double spikes. A directional-mean subtraction filter (D-MSF) detects each CO₂ spike, and final targets are detected by joint analysis of both types of detection. The most important property of CO₂-DS detection is that it generates an extremely low number of false positive caused by background clutter. Only the hot CO₂ spike of a ship plume can penetrate atmosphere, and furthermore, there are only ship CO₂ plume signatures in the double spikes of different spectral bands. Experimental results using midwave Fourier transform infrared (FTIR) in a remote sea environment validate the extreme robustness of the proposed ship-target detection.