Machine learning-based algorithm for circularity analysis

Recently, data collection time for inspection of manufactured parts has been reduced by using machine vision systems that have emerged as a financially viable device, compared to conventional touch-based coordinate measuring machines. However, the overall accuracy and speed of the inspection using the machine vision system is still determined by the ability of fitting algorithms that analyse and evaluate form tolerances from the sampled data. Therefore, in this paper a novel technique that can determine the form tolerances, particularly the circularity, with high speed and accuracy by using support vector machine is explored. The results obtained from the computational experiments show that the proposed technique can be used as a robust fitting algorithm that can ensure high accuracy and speed in the analysis, leading to reduced variations in inspection costs, and eventually improved customer satisfaction by enhanced consistency in the inspection.