Effect of Image Binarization on Drop Diameters Measurement

With reference to agricultural nozzles for pesticide applications, droplet size affects biological efficacy of a phytosanitary treatment, environmental pollution, and operator’s safety. Measuring droplet size spectrum of nozzles is a very complex task which has been addressed in multiple ways over the years. Measurement methods based upon image analysis have to solve the problem of recognizing the droplets and separating them from the background. This is accomplished by means of “binarization” or “segmentation” of the image, and it requires setting a threshold value to separate specific pixel intensities from each other. In this research the effects of the choice of the threshold value on computed volumetric diameters were analyzed. The study was carried out considering images of droplets of an air induction hollow cone nozzle TVI 8002 at the pressure of 1.0 MPa, obtained by applying the liquid immersion method in a custom test bench. A starting threshold value was selected based on the average gray level of the image and then variations of ± 5 units were considered. The results showed that the corresponding variations in volumetric diameters were less than 1%. This means that, once the image acquisition system is established, the binarization process may be made automatic and operator independent, with variations in results fully compatible with the precision of the whole measurement system.