A template-free object motion estimation method for industrial vision system in aligning machine

Abstract

With the popularity of varied sensors as well as the development of communication technologies. Industrial vision systems have been deployed in many manufacture applications. Particularly, industrial vision systems coped with air nozzle are often adopted in aligning machines for object aligning. Since alignment efficiency depends on propriety of blown timing/pressure which is casually manifested in the motion of the observed blown objects, motion estimation method is critical to adopt. Unlike conventional methods typically estimate objects’ motions using prior prepared templates, this paper proposed a template-free object motion estimation method for industrial vision system in aligning machine. By virtue of the properties of industrial images, an observation area with bounding box are initialized for each object, then motion estimation is achieved by updating them following expectation-maximization principle. Experiments revealed that the proposed method is able to achieve continuous motion estimation in less processing time.