Roundness measurements for discontinuous perimeters via machine visions

Fairly extensive research has been made in the area of roundness verifications. However, previous studies concentrate on evaluating roundness errors for parts with complete circles. In real world manufacturing environments, there exist some routine industrial parts with discontinuous perimeters, such as circular parts with keyways and flats or parts with partial arcs. With the advent of low-cost computational hardware, machine vision systems have emerged as a financially viable device in automated manufacturing systems. This paper presents an intelligent heuristic for roundness assessments using machine vision for discontinuous perimeters. The proposed intelligent heuristic incorporates the simulated annealing (SA) and outlier analysis to verify roundness errors according to the ANSI B89.3.1-1972 standard [Measurement of Out-of-Roundness, American Society of Mechanical Engineers, New York, 1972]. Several objects with discontinuous perimeters are used as test examples to verify the viability of the developed vision-based measurement system. Experimental results show the proposed approach is equitable for machine vision applications, and can realize the automation of roundness measurements. (C) 2002 Elsevier Science B.V. All rights reserved.