ベーンポンプ用ベーンの平面部・曲面部上欠陥検出の自動化

概要

論文の詳細を見る
A machine that can detect automatically defects such as flaws, notches and scratches on edges or planar and cylindrical surfaces of vanes has been developed. Most flaws on the edge can be detected using the technique of "oblique illumination and normal detection", but the flaws inclined largely toward the surface of machine parts cannot be detected by the above-mentioned technique. In this paper, we clarify that such flaws with large inclination on the edge, and notches and scratches on the surface can be detected using the technique of illumination and detection normal to the surfaces of vanes. Defect candidates are recognized by binarizing the detected image and the maximum lengths "l" and the areas "a" are counted. "a_0" is calculated in order to eliminate a sampling error which is generated in the binarizing process, where a_0 is 1.3l. Experimental results reveal that the defect candidates with areas larger than max(20,a_0) pixels are true defects.
社団法人精密工学会の論文
2002-12-05