Click on “Download PDF” for the PDF version or on the title for the HTML version.
If you are not an ASABE member or if your employer has not arranged for access to the full-text, Click here for options.
Characterizing the Response of a Hyperspectral Camera Used in Close Range Imaging under Laboratory Conditions
Published by the American Society of Agricultural and Biological Engineers, St. Joseph, Michigan www.asabe.org
Citation: Paper number 131594789, 2013 Kansas City, Missouri, July 21 - July 24, 2013. (doi: http://dx.doi.org/10.13031/aim.20131594789) @2013
Authors: Peter Ako Larbi, Suraj Amatya, Manoj Karkee
Keywords: VNIR hyperspectral camera, diffuse reference standard, image processing, light reflectance, sensitivity analysis
Abstract. The spectral output of a hyperspectral camera is sensitive to both internal camera settings and external environmental conditions. In this study, the global sensitivity of the spectral response of a diffuse reference standard to exposure time, target distance from camera, and intensity of light incident on the target in close range (indoor) imaging was investigated. The imaging setup comprised a visible and near-infrared (VNIR) hyperspectral camera (Headwall Photonics Inc., Fitchburg, MA) fixed to a pan-and-tilt unit (FLIR Systems, Inc., Burlingame, CA) and mounted on a tripod stand, with a pair of halogen lights placed behind the camera to illuminate a target. The target was a diffuse reference standard (Gooch & Housego, Orlando, FL), which is reported by the manufacturer to reflect over 99% of incident light. The setup represented a fixed target and light source scenario. Exposure time (four levels), target distance from camera lens (three levels), and light intensities (four levels) were varied in a complete factorial experiment. Only light intensity, exposure time, and their interaction were significant, contributing to explaining 88% of variability in the peak response (signal intensity) with 12% unexplained variability. Overall, light intensity accounted the most for the variability in the output, followed by exposure time. These results provide an objective explanation to some major sources of uncertainty in hyperspectral images obtained from close range targets. Nevertheless, further knowledge of how much influence is passed on to reflectance (normalized) data is needed in order to achieve more accurate hyperspectral data correction.
(Download PDF) (Export to EndNotes)