Abstract
Abstract Programming a vision inspection cell involves several steps: the first deals with the placement of a camera and lighting sources in the cell, the second is concerned with image acquisition and low level processing, and the third is scene interpretation, consisting of pattern recognition and dimensional control. In previous approaches to the problem these steps were not interconnected, resulting in non-optimal solutions for the vision inspection cell configuration. In this paper we propose an integrated approach to vision programming, based on a CAD workstation and making much use of simulation tools, not only for the geometrical cell modelling, but also for camera placement, kinematic behaviour, image synthesis and processing and model matching. This approach makes it possible to back-reference to different levels in the learning procedure in order to enhance vision inspection, by modifying camera angles and positions, changing image grabbing and processing parameters and adapting image processing and pattern recognition algorithms. By running a complete simulation the operator verifies that his strategy is successful. Numerical results arc provided for each of the modules, demonstrating that off-line integrated vision programming closely resembles the reality of on-line inspection, simplifies a given application and enables the global design of an inspection for optimal configuration on the production line to be carried out.
Published Version
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call