Image data acquisition, processing and object recognition for use in a machine vision system

Abstract

Since the early pioneering work of Rosenfeld and Sklansky, machine vision has been used successfully in a number of practical applications. In this work the areas of application have been reviewed to determine the extent to which computer vision has been utilised in industry. Its application so far has been restricted principally to the mass production field of car manufacture, in the areas of body panel inspection and to a lesser degree to unit assembly. The developnent of robots for picking unsorted components fron containers and for inspection depends upon the reliability and the speed of the vision control system employed. This project investigates a preliminary approach for fast shape recognition of randomly positioned and oriented components. The results obtained indicate that this goal can be achieved by the use of the techniques of contour tracing, Fourier and surface texture analysis and shape fron shading methods.