/AnMtgsAbsts2009.54982 Digital Image Analysis for Determination of Ground Cover in Cover Crop Systems.

Wednesday, November 4, 2009
Convention Center, Exhibit Hall BC, Second Floor

Eric Nord1, Jouke Postma2, William S. Curran1 and David A. Mortensen1, (1)Crop and Soil Sciences, Pennsylvania State Univ., University Park, PA
(2)Horticulture, Pennsylvania State Univ., University Park, PA
Poster Presentation
  • Image Analysis Poster.pdf (6.5 MB)
  • Abstract:

    Ground cover, either from living or dead crop residue, or a combination of the two, can strongly influence light availability at the soil surface and affect weed germination and emergence. As such, ground cover is often measured or estimated in studies of weed emergence. The accessibility of digital photography and computer processing power has made digital image analysis a convenient method for measurement of ground cover. While there are several commercial image analysis software packages available, most do not allow the analysis of color ratios (e.g., the ratio of red to green for a given pixel). This is an important limitation, as color ratios can be very useful in discriminating between ground cover classes. This paper presents a general method for analyzing digital images using a combination of free software packages to determine percent ground cover in cropping systems that incorporate cover crops to enhance weed suppression. We also present as a case-study, analysis of ground cover in two winter-cover crops: cereal rye and hairy vetch, with several establishment and termination dates. For the rye cover crop, later establishment significantly reduced ground cover (by 2.6%), and later termination increased ground cover (by 2.6%). The color classification for these images was quite robust; only 2-5% of pixels had colors that were within a 2% of any of the threshold values. Our analytical approach allowed relatively rapid assessment of ground cover from digital images, and allowed the detection of small differences in percent ground cover associated with differences in timing of cover crop management. We anticipate that this analytical method could be adapted to a wide variety of experiments that require discerning small differences between living and dead plant residues and the soil surface.