A Robust Moving-Object Detecting Method Using Particle Swarm Optimization for a Billet Location Control

This paper presents a robust moving-object detecting method based on particle swarm optimization for billet location control in the heating kiln using background subtraction. There is not a fixed lighting in the heating kiln, and the illumination would change gradually with the change of temperature in the heating kiln. Background subtraction is the most popular and simple detection method used in quickly detecting and tracking moving object from images. However, it would extract false object information as the illumination changes. This paper proposes a novel multi-background images model for detecting the moving billet in the heating kiln using particle swarm optimization. The algorithm extracts a current background image from these multi-representative background images, which extracts from the sensed images data in off-line learning. This billet location control system gets a good control performance in a workshop.