指紋辨識方法的研究已有多年時間，相關應用已逐漸走入日常生活當中。隨著時間演進，指紋系統以及辨認準確率也不斷提升，然而那些辨識方法往往需要使用較複雜的方法去實現。 在此論文中，我們研究簡單型的指紋辨識方法。執行方式為先用印泥蓋出自己的指紋，經由掃描器取得影像後，再以影像處理軟體強化指紋影像，以建立清楚的指紋資料庫。由於考量樣本指紋可能有角度旋轉的辨識問題，因此利用幾何關係的原理，交叉比對影像與資料庫影像，尋找適當的共同參考座標位置，再以二元關聯法來計算兩兩影像間的相關係數（correlation），取係數值最大者為最相似的指紋影像，依此來作為此論文中新型的指紋辨識的方法。 我們發現使用不同的區塊形式來計算相關係數，對於我們的辨認準確度有著相當的差異。因此，本研究也針對此議題作了一併性的探討。 ;Fingerprint identification methods have been studied for many years, and their applications are also widely developed for our daily lives. As time progresses, the recognition accuracy in fingerprint systems is greatly promoted. However, those well-developed identification methods usually require complicated techniques to implement.
In this thesis, we study a simple method for fingerprint identification. We perform the method with first getting the reference fingerprint taken by a stamp pad and storing the image through a scanner. Then, we utilize the imgae processing software to enhance the fingerprint image, and by the same way, to build a fingerprint database for identification. Considering those samples in the database may have different rotation angles degrading the identification accuracy, we find the geometric relationship between the reference image and the database images to calculate a common reference coordinate. Then, the binary data association method is applied to calculate the correlation between two images for finding the optimum match, so the most similar fingerprint images can be determined from the database. This is the new fingerprint identification method proposed in this thesis.
We also find out that using different calculation pattern for correlation can cause substantial difference in recognition accuracy. Therefore, this issue is also included in our study.
