|Park, Bosoon - UGA|
|Mao, Chengye - INST. FOR TECH. DEV.|
Submitted to: Transactions of the ASAE
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: November 11, 2002
Publication Date: February 1, 2003
Citation: Lawrence, K.C., Park, B., Windham, W.R., Mao, C. 2003. Calibration of a pushbroom hyperspectral imaging system for agricultural inspection. [abstract] Transactions of the ASAE. Interpretive Summary: In imaging camera system was developed to detect contamination on poultry carcasses. The imaging system is known as a hyperspectral imaging system, which can measure the light intensity over a range of wavelengths for every dot, or pixel, of a digital image. However, the hyperspectral imaging system is a research tool, which had not been calibrated, and the data collected with it could not be compared with other data. Therefore, a calibration method for this system was developed. The calibration used a spatial remote sensing technique call geometric control point (GCP) calibration to remove optical errors from the system. GCP calibration simplified the calibration process because every pixel in the image did not need its own calibration equation. Next, the GCP corrected image was calibrated to wavelength and distance measurements with separate cubic linear regressions. Finally, standard reference panels were used to calibrate the system to percent reflectance values. Percent reflectance i a unit commonly used to report spectral data. The calibration enabled the hyperspectral imaging system 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 hyperspectral imaging system had been demonstrated. The method consists of a modified geometric control point correction to remove smile and keystone effect from the system and both wavelength and distance calibrations to reduce the wavelength and distance errors to less than 1 pixel value. Next, a pixel-by-pixel percent reflectance calibration nwas performed at all wavelengths with dark current and 99% reflectance calibration-panel measurements, and results were verified with measurements on a certified gradient Spectralon panel with nominal values ranging from 12% to 99%. Results indicate the method is capable of calibrating the hyperspectral system across the entire spectral range of the detector, but errors increase significantly below 420 nm and above 840 nm. Further research should be performed to evaluate the stability of the calibration over time, and techniques must be developed to implement the calibration for real-time analysis.