Combination of simple vision modules for robust real-time motion tracking

摘要: In this paper we describe a real time object tracking system consisting of three modules (motion detection, tracking, robot control), each working with moderate accuracy, implemented in parallel on workstation cluster, and therefore operating fast without any specialized hardware. The robustness quality the is achieved by combination these vision an additional attention module which recognizes errors during tracking. For image sequences apply method active contour models (snakes) can be used for description extraction as well. We show how snake initialized automatically motion detection module, explain demonstrate module. Experiments that approach allows robust real-time over long sequences. Using formal error measurement presented it will shown moving center 90 percent all images.