A Novel Feature Detection Method Using Multi-Dimensional Image Fusion for Automated Optical Inspection on Critical Dimension

適用於關鍵尺寸自動光學檢測的創新多維融合圖像特徵偵測方法

陳亮嘉(Liang-Chia Chen) ； 梁景文(Ching-Wen Liang) ； 黃庭強(Dinh-Cuong Hoang) ； 楊德孝(Duc-Hieu Duong) ； 陳金聖(Chin-Sheng Chen) ； 林世聰(Shyh-Tsong Lin)

《中國機械工程學刊》 39卷2期 (2018/04) Pp. 145-152

摘要

This paper presents a novel approach which is based on multi-dimension image fusion to effective extraction and segmentation of edge features for accurately measuring critical dimension on objects having complicated surface patterns or random reflectance. In the approach, coarse estimation of edge points is firstly performed by using the 3D edge detector to identify correct image regions of interest (ROI) for object segmentation. 2D image processing algorithms are performed on the ROI to segment the precise object edges for critical dimension (CD) measurement. To verify the effectiveness of the strategy, the developed method has been verified through measurement of aerospace composite parts for its edge detection and critical dimension accuracy. The measurement repeatability error of this critical dimension can be kept below 1.1% of the measured CD while the standard deviation can be kept less than 0.137 mm. Experimental results have demonstrated the feasibility and applicability of the developed method.

關鍵字

Automated optical inspection (AOI) ； object segmentation ； image fusion ； point clouds ； critical dimension (CD)

並列摘要

本文提出一個結合三維點雲提取邊緣和二維精確圖像特徵偵測的方法。先以量測件表面深度進行三維邊緣偵測可克服二維影像光強變異及量測面花紋的等困難，三維邊緣偵測結果映射至同步擷取的二維影像，再以二維圖像特徵法則可得到量測件的圖像特徵。航空複合材料工件上的關鍵尺寸量測重複度小於1.1%，証實發展方法具有潛力可以發展而應用於精密製造。