A Swarm Intelligence Approach to Counting Stacked Symmetric Objects

In a manufacturing environment it is often necessary to perform a manual inventory of finished goods and raw materials. These raw materials might be wood, plastic or metal and often represent a large investment of capital to procure and store. While the economic value of an accurate inventory is high, the process of obtaining a good count is tedious and fraught with human error. As a precursor to counting an inventory of tubular steel bar stock from digital images, we present a hybrid algorithm inspired by ant colony and particle swarm technology. This algorithm defines an environmental habitat for interacting particles to competitively cluster into segmented colonies. By forming colonies at bar stock end profile locations the algorithm provides potentially countable high-level information.