A Multiple Background Images Model for Billet Location Control

Abstract

This paper addresses the problem of real-time location of billet in the heating kiln of a steel mill, and presents a vision-based steel billet location control system. The locations of the billets are obtained by background subtraction method. The major difficulty is the strong illumination changes in the heating kiln with the temperature changes. This paper adopts the idea of multiple background images model to adapt the illumination changes in the heating kiln. The multiple background images model classifies the background images as three representative background images, and provides an adaptive weight mechanism for switching a background image among three representative background images. The practical experimental result shows the algorithm is effective and fast, and can meet the demand of real-time control. The vision-based location control system has been successfully used to track the billet edge in industrial condition, and can obtain good control performances under the industrial environment.