Image analysis to quantify herbicide efficacy for Cynodon dactylon control

Abstract

Studies on the effectiveness of herbicides for control of Cynodon dactylon (L.) Pers. (cynodon) normally rely on visual recording methods, which are rapid and convenient, but lack objectiveness and reproducibility. To address this problem, a computer software package was customized for creeping grass studies and applied to determine the ’cover of green’ from digital images of cynodon to verify and interpret visual assessments of herbicide treatments applied to cynodon-infested plots, with or without disking. Whereas the digital analyses confirmed the main conclusions from the visual assessment, significant differences between the recording methods were found during early and late evaluations of the trial, thereby having implications for the interpretation of assessments during these times. These differences were attributed to the difficulty of visual assessments early in the trial when there was partial grass control and late in the trial where seasonal grass senescence was superimposed to the treatments. The advantages of digital analysis are ease of use, repeatability, accuracy and objectivity enabling more profound analyses; as well as reduced costs, although in this study, sample number and size was restricted due to labour-intensive maintenance of the target species.