影像處理在圖形辨識與機械手臂裝配應用之研究

The Study of Image Recognition in Engineering Drawing and to Robot Arm Assembly

王中行(Chung-Shing Wang) ； 李昭龍(Jau-Long Lee)

《大葉學報》 6卷1期 (1997/12) Pp. 101-109

https://doi.org/10.7119/JDYU.199712.0101

摘要

在自動化的發展上，影像辨識與產業自動化關係非常密切，辨識物體功能的高低，直接影響到產能。本文以影像辨識為基礎，分成三階段的研究，第一階段為手繪工程圖的辨識，發展出一工程圖辨識系統，將工程圖經掃瞄後可直接轉至AutoCAD軟體，以便保存與修改。第二階段為發展3D立體物之辨識程式，利用三個或單一攝影機對於物體進行辨識，並將其轉成實體模型。第三階段為影像辨識與機械手臂配合，以物體分類及軸孔裝配為其動作之分析。本研究著重於應用性，已完成2D手繪工程圖之辨識軟體、3D物體之辨識軟體、具視覺之機械手臂分類程式、軸孔裝配之程式等成果，均可提供產業界利用。

關鍵字

2D手繪工程圖辨識 ； 3D物體辨識 ； 實體模型 ； 機械手臂裝配

並列摘要

Image recognition and manufacturing automation have a very close relationship in the development of industrial automation environment. The capacity of image recognition can greatly affect the quality and quantity of products. This research based on image processing and its application. Three goals have been achieved. First goal is free-hand engineering drawing recognition and recreation. This research has accomplished by receiving the bitmap format data from scanning a free-hand drawing and transferring to vector format data in AutoCAD software for easy store and modification. Second goal is the development forward to three-dimensional object recognition. By using one or three cameras, the objects image pattern can be rebuilt into the solid model of AutoCAD software. Third goal is to implement the use of image recognition to robot arm assembly. Object assortment and axis-hole matches are two ways in assembly process. This research is based on practicality. The achievements included 2D free-hand engineering drawing recognition and recreation, 3D-object pattern rebuild and the objects assortment and axis-hole matched with robot arm. The whole results can provide a great use in industry.

並列關鍵字

2D free hand drawing, image processing ； 3D image recognition ； solid model ； robot arm assembly