Skip to Main Content
Summary form only given, as follows. The passage of a surface ship through a body of water produces a characteristic thermal scar which manifests itself as a radiant intensity contrast pattt????????????m. This pattern is visible in the thermal infrared band (8-14 microns) and may persist for a number of kilometers. Recent advances in IR detector array technology have enabled new infrared sensor systems to be developed that can image these thermal scars with spatial resolutions and temperature sensitivites previously unattainable. The level of IR ship wake detail obtained with these advanced IR sensors provides new opportunities for developing and testing robust detection algorithms. A modem state-of-the-art, calibrated FLIR is described that was developed for imaging low contrast ocean backgrounds. Particular attention is paid to the calibration methods employed to reduce pattern noise and other systematic errors down to theorectical levels. Two robust algorithms for the automatic detection and discrimination of thermal infrared surface wakes are presented and compared to a classical energy detection approach. Both approaches exploit contrast and textural differences between the signal and ambient background clutter. The detection algorithms were implemented on actual imagery of a mono hull sailboat wake obtained with the digital FLIR described above, and were also tested on simulated wake imagery. The relative performance of these detection schemes versus c:onventional approaches is quantitatively assessed via ROC (Receiver Operating Characteristic) curves, and comparisons made of derived detection statistics and metrics. Particular attention is paid to the the problem of detecting and discriminating very low contrast (< 0.05 degrees Celsius) signatures utilizing robust detection schemes in conjunction with high-resolution, high-sensitivity IR imaging.