Evaluation of Key Methodology for Digital Image Analysis of Turfgrass Color Using Open‐Source Software

Abstract

Digital image analysis is a frequently used research technique to provide an objective measure of turfgrass color, in addition to the traditional visual rating. A commonly used method relies on commercial software package SigmaScan Pro to quantify mean hue angle, saturation, and brightness values from turf images, and to calculate a dark green color index as the measure of color. To enable turf image analysis to function on an open‐source platform, a method was developed within ImageJ to batch process turf images for color parameters. This Java‐based ImageJ plugin quantifies hue angle, saturation, and brightness values and calculates a dark green color index. In addition, information on the variability of these color parameters can be simultaneously acquired. This new method was used to quantify color parameters of turf images collected from field plots of tall fescue (Schedonorus arundinacea Shreb. Dumort.), Kentucky bluegrass (Poa pratensis L.), ryegrass (Lolium ssp.), hybrid bermudagrass (Cynodon dactylon (L.) Pers. × C. transvaalensis Burtt‐Davy), and creeping bentgrass (Agrostis stolonifera L.). While color parameter values differed little between ImageJ and SigmaScan, the time saved in processing images using ImageJ was considerable. Aside from software, analysis of color parameters acquired from the five turfgrass species indicated that hue angle alone can adequately measure turf color in digital images. Results also demonstrated that, in addition to light source, camera settings should remain fixed during photo capture to avoid introducing errors. The ImageJ plug‐in developed in this study is made available at http://www.turffiles.ncsu.edu.

Sewage sludge composted in the coloring and development of Bermuda grass

“…Thus, experiments with digital image analysis are efficient to estimate the nutritional status of the lawn (Zhang et al, 2017), especially with regard to nitrogen, and corroborate the studies that correlate this analysis with nitrogen doses by Backes et al (2010), Lima et al (2012), Agati et al (2015) and Gazola et al (2016).…”

Color‐distance modeling improves differentiation of colors in digital images of hybrid bermudagrass

“…Zhang et al. (2017) reported that θ generally outperformed saturation or brightness when used to identify turfgrass color variations among genotypes or due to fertility levels. However, the current research suggests concurrent analysis of saturation and brightness (or analogous components) along with θ may be required to obtain a more complete picture of color.…”

“…The same is true for DGCI. While DGCI was effective in assessing turf colors (Zhang et al., 2017) and does include aspects of saturation and brightness, it may not be suitable for examining colors other than green, which may preclude follow‐up assessment of shoots treated with herbicides or growth regulators or other chemicals or cultural practices known to elicit phytotoxicity reducing green color.…”

“…Specifying color in digital images of turfgrass is straightforward (Berndt, Riger, & Riger, 2018a; Karcher & Richardson, 2013; Zhang, Pinnix, Zhang, Miller, & Rufty, 2017) but reporting of resulting color space data may be improved by differentiating colors in photographs more accurately. Having the ability to accurately differentiate colors is important because color differentiation has implications for many facets of turfgrass research involving pesticides, fertilizers, colorants, cultivars or species, and/or other areas.…”

“…Having the ability to accurately differentiate colors is important because color differentiation has implications for many facets of turfgrass research involving pesticides, fertilizers, colorants, cultivars or species, and/or other areas. Statistical analysis of hue angle (θ) and a hue, saturation, brightness (HSB) color space derivative, termed dark green color index (DGCI), have been a basis for identifying and reporting color differences in turfgrass research (Karcher & Richardson, 2013; Zhang et al., 2017).…”