016 Characteristic Dimension: A Novel Adjunct to Analyzing Species Composition in Digitized Photos of Turfgrass Plots

in HortScience
View More View Less
  • 1 Kansas State University, Dept. HFRR, Manhattan KS 66506

Digitized photographic images of turf plots composed of bermudagrass, buffalo grass, tall fescue, and zoysiagrass were taken at a height of about 150 cm with a 28-mm lens. Fast Fourier transforms of these images were performed, and a radial plot of the power spectrum was obtained from each image. Hurst plots (log frequency vs. log intensity) were used to subtract “background” from the power spectra, so peaks would be more evident. The peak of the power spectrum occurs at the average spacing between leaves (more precisely, between areas of the canopy that reflects a significant amount of light) and defines the characteristic dimension. Zoysiagrass had the lowest characteristic dimension, while tall fescue had the highest. The width of the power spectrum is indicative of the variability of the characteristic dimension within the canopy. The minimum characteristic dimension (occurring at the highest frequency) was less than 1.7 cm, whereas all the other species had about the same minimum characteristic dimension of ≈1.9 cm. The maximum characteristic dimension was greatest for fescue (6.9 cm), followed by buffalo grass (3.8 cm), bermudagrass (3.3 cm), and zoysiagrass (2.8 cm). These results indicate that the characteristic dimension can be a useful tool for discriminating between turfgrass species in digitized images.