Development of an Image-Processing Method to Assess Spreader Performance

Abstract

Fertilizer spreader testing plays an important part in achieving accurate, uniform application of nutrients to the land; however, current testing methods are often laborious, with high levels of manual input required to obtain specific spread characteristic information. This study aimed to review and identify potential techniques and test one that could be incorporated into a spreader testing program. A review of available spreader tests throughout the world showed varying results. Measurement systems requiring a mathematical algorithm to calculate the landing position of individual particles were found to be quick interpolators, but inaccurate. Tray collection, used extensively by quality assurance programs throughout the world, is accurate in producing distribution pattern results; however, it is a laborious task, especially when analyzing multiple spread patterns for variable-rate application. An image-processing method showed the greatest potential as a rapid analysis and interpretation tool of spread pattern data. A computer program was written to analyze and extract individual particle information from urea fertilizer contained on collector trays during a spreader test. Results indicated that there was a strong relationship between two-dimensional particle area and particle mass under laboratory (R2 = 0.991) and field (R2 = 0.988) test conditions. Measurements taken within 4 m of the spreading vehicle had a higher expected error when predicting particle mass, which was attributed to a high proportion of particles with a cross-sectional area less than 1 mm2. High levels of expected error in field results indicated that particles need to be greater than 1 mm2 after spreading if the method was to accurately predict an entire distribution pattern of a fertilizer spreader. However, the method was found to accurately calculate the size guide number and uniformity index of the collected sample.