2208 FRP パネルの外観検査画像情報処理の自動化と欠陥分布評価

概要

論文の詳細を見る
This paper deals with a digital image analysis for automated visual inspection methodology for detecting the surface defects in FRP products in order to enhance human skill with machine performance for improved management of product flexibility and product quality. We studied the color image processing with HSV (Hue, Saturation, and Value of lightness) color space. An external appearance of FRP products is transformed into the original appearance image data captured by CCD camera. The proposed image processing consists of threshold method, noise elimination, edging, and elliptic approximation on principal axis of intertia. The automated threshold method is based on maximum variance method. The elliptic approximation on pricipal axis of inertia generates the simple shape in stead of the original defects from the appearance image data. In addition, the new methods of evaluating distribution and classification of surface defects was also developed.
社団法人日本機械学会の論文
2001-01-16