Tracking capabilities in computer vision for industrial applications

Abstract

A tracking system has been developed with a compromise between complexity and cost and efficiency in mind. The tracking system consists of a PC fitted with a specific one-card processor designed in our laboratory that acquires the image and implements the recognition algorithm at video rate. We describe the recognition method, the hardware architecture, and some applications and results. The system can be extended to track targets in 3-D using two cameras and two processors. Finally we describe different applications already tested: A robot servo-positioning system in moving parts operation, a tracking system for vehicles for traffic supervision and control, and a tracking system for people for the study of their behavior.