在自動化的視覺工業檢測系統中,影像處理技術一直左右著檢測系統的好壞。此篇論文中,針對X-Ray所拍攝到的多層印刷電路圖,以全自動的方式找出所有的圓孔資訊,並且以提出的模組化主動式輪廓偵測法增加擷取後結果的精確度。整個演算的流程主要可分為兩個部分:首先是對於印刷電路進行分析,找出可能為圓形的位置資訊,作為主動式輪廓偵測法的初始模組;然後將找出的初始模組,套入所提出的模組化主動式輪廓偵測法以增加其結果的精確度。在實驗分析中,雜訊20﹪以下的干擾,我們的方法可以有0.5 pixel的平均誤差。實際的 X-Ray影像之實驗結果已證實我們所提出之方法的可行性。最後並討論此方法的限制和未來發展的趨勢。
In the automatic optical inspection system, the performance is influenced by the adopted technology of image processing. In this thesis, we present a method to acquire whole information about circular holes automatically from the X-Ray image of a multi-layout printed circuit board(PCB), and design a modular active contour to constraint the result close to the object. Two main stages desired in our approach include the segmentation of probably circular holes information and the accurate location of using modular active contour. The probably circular hole information are obtained by the analysis from the X-Ray image. Then, the probably circular hole information support the initial settings used in developed the modular active contour. According to the initial settings, modular active contour can carry out and push the initial model to locate closely the current circular hole. Experimental analyses have shown that the proposed approach obtains the performance of average error 0.5 pixel under 20% below random noise added. The experiments of real PCB images have also confirmed the feasibility of the proposed approach. Finally, the limitations and future works of the proposed method will also be discussed.