Computational Metrology for Measuring Industrial Component Dimensions

Abstract

To target the problem of dimension measurement of objects in industry, a new computer vision method is proposed based upon Harris-corner detection. The proposed research delivers an alternative to the requirement of various precise measurement devices and skilled labour. In order to measure the various dimensions of an object, the proposed algorithm separates the corner points from the background based on variations in pixel intensity. An algorithm has been proposed to analyze captured object images and perform measurements and inspection processes. The aim of this paper is to utilize computer vision detection algorithms to control the quality of manufactured parts by sorting them on size tolerance. The length of various objects such as screws, bolts, and a rectangular iron piece was determined from the images captured using smartphone camera (Samsung Galaxy F62). The evidence for a total of eight different measurements is presented, and the accuracy of the method is proved up to 99 percent against the dimensions measured using the Vernier calliper.