生物特徵識別（Biometric identification）是目前熱門研究的主題之一，其中有一部份運用手部的特徵像是指紋（Fingerprint）、手的幾何（Hand geometry）、掌紋（Palmprint）與掌背血管（Vein of Palm-dorsum）等。其中，掌紋的特徵由主線（Principal lines）、皺褶（Wrinkes）與脊（ridges）所組成，而人類的整根手指上的紋路與手掌上的紋路亦有近似的三項特徵結構，即主線、皺褶、脊。因此，本論文嘗試取出手掌影像中的小指、無名指、中指和食指共四根手指上中央部位的長方形手指紋路（Finger Crease Patterns）當作興趣區域（Regions of Interest, ROI），再以小波轉換（Wavelet transformation）來取得小波能量特徵（Wavelet Energy Features, WEF），最後藉由倒傳遞類神經網路（Back propagation neural network, BPNN）來作比對。 Biometric identification is one of the popular research fields recently. Some of those use the features of hands like fingerprint, hand geometry, palmprint and vein of palm-dorsum. Palmprint has several features, which include principal lines, wrinkles and ridges. Also, the whole stick of finger is full of the similar structure of features (i.e. principal lines, wrinkles and ridges). In this dissertation, a new approach is introduced. In the image of the palm, we try to use Finger Crease Patterns on the central area of four fingers (little finger, ring finger, middle finger and forefinger) as regions of Interest (ROI). Next we computed the wavelet energy feature (WEF) through the use of wavelet transform. Finally, the back propagation neural network (BPNN) is applied for verification.
