Characterization of 3-5 micron thermal imagers and analysis of narrow band images

• Main Author: Quek, Yew Sing
• Other Authors: Cooper, Alfred W.
• Format: Others
• Published: Monterey California. Naval Postgraduate School Dece
• Subjects: Imaging systems; Image analysis; Thermal imager; Narrow band images; Narrow band filter; Minimum resolvable temperature; Noise equivalent temperature difference; Modulation transfer function
• Online Access: http://hdl.handle.net/10945/1237

Approved for public release, distribution is unlimited === In this study, the use of multi narrow-band filters was explored to correlate or fuse information to improve detection and recognition for specific targets of interest. To fully understand the viability and limitations of narrow band filters in thermal imaging, laboratory test procedures were first designed to characterize thermal images using common performance parameters, particularly the Noise Equivalent Temperature Difference (NETD), the Modulation Transfer Function (MTF) and the Minimum Resolvable Temperature (MRT). An available thermal imager, the Cincinnati Electronics IRRIS-256LN, and a newly purchased thermal imager, the Indigo Systems Merlin InSb Laboratory Camera, were investigated and compared. The MRT measurement showed a superior cut-off spatial frequency of 1.33 cycle/mrad for the Merlin camera, compared with 1.18 cycle/mrad for the Cincinnati, with similar sensitivity of about 0.1 ʻC for both cameras. Below cut-off the Modulation Transfer for the Merlin was approximately twice that of the Cincinnati. NETD measurements for both systems were inhibited by excess noise in the measurement system. The Merlin camera was consequently employed for further research on narrow band images. Three commercially available narrow band filters, 4110/4720 nm, 4540/4720 nm and 3750/4020 nm, were selected to isolate the "red" and "blue" spike signatures of plume emission and aerodynamic heating found in the signature of fast moving air targets. Similar sensor characterization was carried out with these filters, however, quantitative performance parameter measurements were hindered by excessive measurement system noise. As a preliminary study in multi-spectral image analysis, spectral features of hydrocarbon combustion were extracted from correlated narrow-band images of a laboratory propane flame. Digital analysis using software statistical tools showed potential for discrimination of hydrocarbon emission from other thermal sources by correlating pixel number and pixel radiant intensity. === Captain, Singapore Armed Forces