印刷電路板在電子資訊工業佔著舉足輕重的地位，世界各主要工業國家均列為最重要的基礎工業。雖然電路板組裝的過程已經是十分自動化了，但仍然有許多檢測工作需要檢測員目測，這樣做法不僅浪費時間、而且檢測員會因個人主觀不同或者視覺產生疲勞，使得檢測結果不一致，讓品質控管變的困難。本論文主要研究以電腦視覺檢測技術為主的系統來替代人工目視的方式，針對電阻、電容及二極體等電路板元件做缺件、錯件及極性相反等檢測。此系統首先利用二值化(thresholding)、邊緣偵測(edge detection)及細線化(thinning)等基本影像處理對電路板進行自動定位並找出待側元件的位置區塊，然後以差異值(variance)和不變矩(invariantmoment)配合歐幾里德距離對待測元件進行辨識。最後以實際電路板元件進行實驗，其結果顯示出98%以上的整體辨識率，驗證了本論文所提電路板元件之視覺檢測方法的可行性及正確性。Printed Circuit Board (PCB) plays a very important role on electronic informationindustry. It has been listed as the most important and fundamental industry in manyindustrialized countries around the world. Although the assembly of the circuit board has beenmore automatic than it used to be, there are still much inspection work to be done by theinspector with their eyes, which not only wasting time, but also getting inconsistent resultsdue to different judgment of the inspectors or visual tiredness, so that the quality controlbecomes difficulty.This paper mainly renders the idea to use the system based on computer visualtechnique instead of manual visual inspection method. The inspection of our system for thecircuit board elements such as missing item, wrong item and reversed polarity, etc., has aimedat electric resistor, capacitor and diode. This system first used the basic image processingmethods including thresholding, edge detection and thinning, etc., to proceed with autopositioning on circuit board. Then we used the variance and invariant moment to coordinatewith Euclidean distance to identify the elements to be inspected. Finally, this thesis used thereal circuit board elements to do experiment. The results reveal the whole recognition rateover 98% that verified the feasibility and accuracy of the visual inspection method submittedby this research for PCB elements.