REFLECTANCE CALIBRATION OF FOCAL PLANE ARRAY HYPERSPECTRAL IMAGING SYSTEM FOR AGRICULTURAL AND FOOD SAFETY APPLICATIONS

Authors: Lawrence, Kurt; Park, Bosoon; Windham, William; MAYO, CHENGYE - GATAN, INC; Poole, Gavin

Submitted to: Society of Photo-Optical Instrumentation Engineers
Publication Type: Proceedings
Publication Acceptance Date: 9/23/2002
Publication Date: 9/23/2002
Citation: Lawrence, K.C., Park, B., Windham, W.R., Mayo, C., Poole, G.H. 2002. Reflectance calibration of focal plane array hyperspectral imaging system for agricultural and food safety applications. Society of Photo-Optical Instrumentation Engineers.

Interpretive Summary: Remote sensing is a research field that has been using imaging systems operating over large wavelength regions for several decades. Most systems, known as hyperspectral imaging systems (HIS), collect both full spatial information (digital camera) and spectral information (spectrometer) for each and every pixel of an image. Cost of HIS are decreasing and more small research teams are now using them. Often, there is not a dedicated staff to calibrate the systems and many, who are unfamiliar with the complexities of hyperspectral calibration, are using the systems uncalibrated. Furthermore, HIS are being applied to new and innovative problems like food safety applications and near-field agricultural applications. Although researchers can usually show proof of concept with uncalibrated systems, they cannot compare their results with others because their systems don't have a common reference. This paper provides a simplified method to calibrate pushbroom HIS. The calibration uses geometric control points to remove optical errors in the system and enables the HIS to measure percent reflectance with an error of less than 5 percent difference over the range from 420 to 840 nm. Thus, the results from the system can now be compared with other imaging systems and techniques.

Technical Abstract: A method to calibrate a pushbroom hyperspectral imaging system for "near-field" applications in agricultural and food safety has been demonstrated. The method consists of a modified geometric control point correction applied to a focal plane array to remove smile and keystone distortion from the system. Once a FPA correction was applied, single wavelength and distance calibrations were used to describe all points on the FPA. Finally, a percent reflectance calibration, applied on a pixel-by-pixel basis, was used for accurate measurements for the hyperspectral imaging system. The method was demonstrated with a stationary prism-grating-prism, pushbroom hyperspectral imaging system. For the system described, wavelength and distance calibrations were used to reduce the wavelength errors to <0.5 nm and distance errors to <0.01mm (across the entrance slit width). The pixel-by-pixel percent reflectance calibration, which was performed at all wavelengths with dark current and 99% reflectance calibration-panel measurements, was verified with measurements on a certified gradient Spectralon panel with values ranging from about 14% reflectance to 99% reflectance with errors generally less than 5% at the mid-wavelength measurements. Results from the calibration method, indicate the hyperspectral imaging system has a usable range between 420 nm and 840 nm. Outside this range, errors increase significantly.