Use of Digital Image Analysis for Counting Dandelion Blooms in Field Plots.

Digital image analysis provides researchers a method to accurately and efficiently analyze turfgrass parameters, which include percent coverage, color, and disease severity. Recently, ImageJ (National Institutes of Health, Bethesda, MD) has replaced SigmaScan (Systat Software, Inc., San Jose, CA) as the image analysis software of choice, mainly due to the free and open-source nature of ImageJ. We developed an ImageJ macro that is able to count the number and quantify the percent coverage of dandelion blooms in images of field plots, which can assist in broadleaf weed control data collection. Two commands were coded into the macro to separate the individual dandelion blooms from interfering objects. First, the particle analysis function was used to distinguish dandelion blooms from other yellow objects (e.g. chlorotic turfgrass leaves) based on their size and circularity. Next, binary watershed segmentation was used to separate groupings of dandelion blooms into individual blooms that could be counted. To verify the accuracy of the macro, we analyzed 168 images from broadleaf weed control studies for the number of dandelion blooms with the macro and regressed it against our visual counts. The resulting regression trend line (visual count vs. macro) had a slope of 1.02 and an R² value of 0.976. Dandelion bloom counts ranged from 0 to 261 and 0 to 250 for the macro and our visual counts, respectively. This macro provides researchers with a rapid and accurate method of determining the number and percent coverage of dandelion blooms in field plots. Furthermore, this macro shows that it is possible to quantify weed coverage/encroachment via image analysis in certain situations.