DEVELOPMENT OF SYNTHETIC IMAGERY FOR VISUALIZATION OF COTTON REFLECTANCE

Authors: Sassenrath, Gretchen; ALARCON, V - MISSISSIPPI STATE UNIV.

Submitted to: American Society for Photogrammetry and Remote Sensing Proceedings
Publication Type: Proceedings
Publication Acceptance Date: 10/5/2005
Publication Date: 10/6/2005
Citation: Sassenrath, G.F., Alarcon, V.J. 2005. Development of synthetic imagery for visualization of cotton reflectance. 20th Biennial Workshop on Aerial Photography, Videography, and High Resolution Digital Imagery for Resource Assessment, #0038, 4 pgs. American Society for Photogrammetry and Remote Sensing Proceedings. CD-ROM

Interpretive Summary: Remote sensing systems generate large data sets from measured hyperspectral reflectance spectrum. Analysis and interpretation of these data sets is hindered by the quantity of data available. We have developed methods of simplifying the large data sets, without loosing any of the information contained in the data. The process of Fourier transformation allows condensing the information from 512 bands to 20 pairs of harmonics. These reduced data sets can then be manipulated and analyzed in ways to extract the most significant components. Visualization of the reflectance spectra is aided by a process of synthetic image generation. These synthetic images look very similar to remote images containing much less information. The analytical and visualization tools are powerful methods for robust analysis and interpretation of the reflectance data.

Technical Abstract: Remote imagery offers the potential for rapid, timely, and continuous monitoring of crop status, and has the potential to offer significant benefit to crop production and management. However, errors are introduced in moving from the well-characterized measurements of individual leaves to interpretation of factors operating in an entire production canopy. ARS scientists, in collaboration with University colleagues, have developed a methodology of building synthetic images from reflectance spectra of individual leaves. The individual factors impacting canopy reflectance can then be examined separately. The method initially approximates leaf spectral characteristics through the use of Fourier transformations that adequately describe the spectra and reduces the information required from 512 bands to 20 pairs of harmonics. These spectra from soil and vegetation are then randomly mixed and visualized using an imaging software package. The resultant synthetic spectra are remarkably similar to actual remote images of crop fields obtained from aerial sources. This methodology offers a powerful analytical technique to explore crop reflectance spectra, and build spectral libraries of reflectance patterns with physiological and morphological changes.