利用影像處理技術於印刷電路板線路追蹤

Abstract

在印刷電路板線路設計時，工程師通常要以人工方式比對線路，而伴隨著半導體科技的進步，印刷電路板的線路越來越複雜，單靠人眼已無法達成快速檢查的要求。所以希望能以影像處理的技術來進行全自動的分析檢查，來降低失誤率及所需的時間。 本論文以影像處理技術來達到線路追蹤的目的，以不同的顏色搭配影像標籤(labeling)演算法來標示線路，再以圖像辨識的方法以及形態學處理方法來做鑽孔層的連結分析，這樣我們就能知道各層線路的連結情形。最後再以SUSAN轉角偵測法來找出各類零件的連結關係。 提出的方法已經可以很正確的將各層的線路正確地連結起來，也能辨識出電阻及電容打印的部份。

Engineers usually compare circuit layout schematics by themselves in design stage. However, with the semiconductor technology improvement, the more and more complex PCB design has caused people unable to quickly check the layout by human eyes. In this thesis, we try to analyze the schematic automatically by image processing technology to reduce the mistakes and save a lot of time. To elaborate, this study applies the image processing technology to complete the layout tracking, distinguish each line by different colors with labeling algorithm, and analyze drill type by pattern recognition and morphological operations. In the end, we can get the connectivity of each layer of the layout image, and get the relationship of components by using SUSAN algorithm to recognize the footprints. Test result show that our method can track the netlist through every layer correctly. It can also recognize basic components like resistors and capacitors.