在近年來指紋被當作是生物識別的一個重要工具,而犯罪現場的重疊指紋辨識是警方辦案的困擾。因重疊與未重疊指紋邊界區域的方向場計算,易受重疊指紋區域影響,導致錯誤的指紋分離與辨識,本論文提出重疊指紋自動分離系統及改良式指紋方向場增強演算法以利辨識。首先,進行重疊與非重疊區域的判別。本文利用重疊區域有兩個方向場且指紋山脊點數較未重疊區域多的特性進行初步分離。因核心點區域與重疊區域有類似特性,容易被誤判為重疊區域。本文再利用核心點的偵測,修正重疊與未重疊區域。最後,利用指紋輪廓明顯與否區分兩枚未重疊指紋區域。接著進行兩枚指紋的分離與辨識。因在重疊與未重疊指紋邊界區域的方向場計算容易產生錯誤,本文提出遞迴式的方向場演算法,再結合限制鬆弛標籤演算法分配方向場[11],最後用Gabor-filter進行分離與辨識[5]。經實驗證實,本系統與手動指紋分離接近。 Overlapping fingerprint at the crime scene the police handling of the plagued. The direction of the field do not overlap as the correct information to separate the overlapping area. Overlapping and not overlapping border area fingerprint orientation field calculation is easy to mistake. This paper presents overlapping separation system and improved automatic fingerprint orientation field enhancement algorithms in order to facilitate identification. Preliminary estimates suggest that the use of the overlapping area and fix by core point and tour, In this paper, proposed the direction of the field recursive algorithm, combined with restrictions relaxation labeling algorithms dispensing direction field [11], and finally with Gabor-filter separation and identification [5]