The indented cylinder is used for sorting seeds. The seeds are physically manipulated in a way, such that sorting is based primarily on the length of individual seeds. Seeds are separated into at least two subsets: (1) seeds of a length a threshold (a scalar) and (2) seeds of a length the same threshold. The threshold is dependent on a number of different parameters. Besides the seed length, the rotation, general size, shape, and surface texture of each seed, are also known to influence the final sorting result. Such knowledge comes from previous experimentation with the indented cylinder. In our work we will seek to understand more about the internal dynamics of the indented cylinder. We will apply image analysis to observe the movement of seeds in the indented cylinder. This work is laying the groundwork for future studies into the application of image analysis as a tool for autonomous inspection in seed cleaning equipment. A prototype of an indented cylinder will be constructed. To make it more dynamic, the cylinder itself will be manufactured using 3D printing technology. The input will come either from 3D scans of existing cylinders or by defining their topology using parametric B-Spline surfaces. Using image analysis, the seeds will be tracked using a kalman filter and the 2D trajectory, length, velocity, weight, and rotation will be sampled. We expect a high correspondence between seed length and certain spatially optimal seed trajectories. This work is done in collaboration with Westrup A/S, Denmark.
Seed Symposium Abstracts, 29th Ista Congress Cologne, 2010, p. 15-16