Laboratory Evaluation of Seed Metering Device Using Image Processing Method

Abstract

Seed-metering system is an important component in row-crop planters in terms of uniform seed distribution. Numerous field and laboratory methods have been developed and used for evaluation of planter performance; each method having its own advantages and shortcomings. In the present study, a digital camera (Nikon, D70) was used for laboratory evaluation of vertical-rotor seed-metering device performance, indices such as multiple planting, feeding quality, miss planting and seed space uniformity being the major criteria. To validate the results from image processing method, the more conventional, grease-belt method, was used. The experiments were conducted on the basis of factorial randomized complete design with two types of seed-metering devices, with different numbers of cells and four levels of seed-metering speeds in three replications. Captured images were transferred to the computer via USB port and were processed by program written in MATLAB software environment. The results from two methods, namely, image processing and grease belt methods were in good agreement and analysis of variances showed that seed-metering device with 15 cells in 40 rpm performed better than seed-metering device with 21 cells and in other speeds. Comparison of the number of spaces that fell in normal domain in different seed-metering speeds showed that image processing method had higher value than grease-belt method. Since the initial falling velocity of seed and angle of its exit from metering device was unknown, thus the precision of space index was different in two methods.