Operational Image Processing Approach for Spray Coverage Measurements in Pesticide Application

Abstract

Water sensitive papers (WSPs) are mostly used to determine the performance of spray systems used in pesticide application. The most critical stage of an image processing operation is determining the threshold level of WSP images. In this study, an analysis of WSP images was conducted using two thresholding methods. The first was a variable-defined approach (dependent), which determined the thresholding level according to the greyscale level of each WSP sample. The second approach was automated, where the thresholding level was automatically set by image processing software. Using both thresholding methods, the percent spray coverage of the WSP samples was determined using a macro module prepared using the software interface, and the results of both methods were compared. Overall, the spray coverage was high and measurement error was clearly observable for some of the WSP images using the automatic thresholding approach. The percent spray coverage of 17 out of 432 analysed WSP images was found to be 24 to 414 times higher when using the automatic as opposed to the dependent thresholding approach. In addition, extreme values for the percent spray coverage was observed for the automatic method. Our results suggest that percent spray coverage without extreme data can be quickly and practically determined for WSP images using a dependent, variable-defined thresholding method using a macro module.