A Study on the Measurement of Mechanical Parts Produced by Wire-EDM Using Machine Vision

• Main Authors: Lian-Shun Chang, 張良舜
• Other Authors: Mu-Tian Yan
• Format: Others
• Language: zh-TW
• Published: 2006
• Online Access: http://ndltd.ncl.edu.tw/handle/80569094794127392659

碩士 === 華梵大學 === 機電工程研究所 === 94 === Vague edge-areas often appear in CCD images of micro-mechanical parts (MMP’s). Possible causes are MMP thickness, high magnification lens, and un-optimized image taken conditions. Difficulties in focusing high thickness of MMP’s using high magnification lens introduce vague edge-areas, and hence, measurement errors. In this paper, a novel measurement method based on a machine vision system is proposed to measure profiles of micro-mechanical parts (MMP) by wire-EDM. In this method, measurement procedures are suggested which extract vague edge-area first, and then, fit and link break-profiles to achieve a more accurate MMP profile. Sobel filter and image processing techniques are employed to segment vague edge-area of a MMP image. Central moment and polynomial curve fitting are then used to find segment-lines in vague edge-area. Finally, Bezier curve is employed to link break-profiles of edges of the MMP image. Experimental results show good accuracy, and can effectively estimate the profile of a MMP image. Indeed, this method is immediately available for industrial applications in MMP measurements. This paper presents a new method that can define profiles for indistinct edges of MMP images by Sobel filter, image processing techniques, central moment and polynomial curve fitting, and Bezier curve. The purposes of these three techniques are: • Sobel, Prewitt filters, and other image processing techniques – to extract inlier profile, suburb profile, and vague edge-area of MMP image. • Central moment and Polynomial curve fitting – to estimate break-profiles of MMP images. • Bezier curve –to joint break-profiles and to obtain an optimal profile of MMP. The objective of this paper is to propose a new method that can establish an optimal profile of a MMP image in order to improve manufacturing accuracy.