Pada asalnya, sistem biometrik adalah sistem pengecaman corak yang beroperasi dengan berpandukan kesahihan anatomi tertentu ataupun ciri-ciri tingkah laku yang dimiliki oleh sesetengah pengguna. Dalam konteks teknologi informasi, biometrik merujuk kepada teknologi yang mengukur dan menganalisis ciri-ciri tubuh badan seperti DNA, cap jari tangan, retina, iris, corak suara, corak muka dan geometri tangan untuk tujuan pengesahan identiti. Pelaksanaan satu sistem pengesahan biometrik yang berpandukan kepada pengecaman tapak tangan akan dibentangkan dalam kertas kerja berikut. Kaedah klasifikasi UMACE dalam erti lain iaitu minima purata tenaga korelasi tanpa kekangan merupakan pendekatan yang dicadangkan untuk pengecaman ciri-ciri tapak tangan. Tapak tangan mengandungi ciri-ciri fisiologi yang penting akan diekstrakkan dengan menggunakan kaedah ROI. Seterusnya, kaedah UMACE akan memainkan peranan penting dalam modul corak padanan. Graf GAR (Kadar Penerimaan Keaslian) lawan FAR (Kadar Penerimaan Kepalsuan) akan diplotkan supaya prestasi sistem biometrik dapat dibandingkan menerusi graf lengkung ROC (Penerima ciri-ciri Operasi). Untuk kesemua latihan data, GAR hanya mencapai 100% hanya pada kedudukan 73% FAR.
_______________________________________________________________________________________________________
A biometric system is essentially a pattern recognition system which recognizes an authentic user by determining the authenticity of a specific anatomical (e.g. palm print) or behavioral (e.g. voice) characteristic or traits possessed by the user. In information technology, biometrics normally refers to technologies that measure and analyze human body characteristics, for example DNA, fingerprints, palm prints, eye retinas, irises, voice patterns, facial patterns and hand geometry, for authentication purposes. In this paper, an implementation of an individual verification system will be addressed based on the palm print recognition. UMACE filter or in other words unconstrained minimum average correlation energy is the proposed approach for the palm print feature recognition. For the feature extraction, the palm print containing the important physiological traits will be captured and extracted into the ROI region of interest. Next, the pattern matching will involve UMACE filter as the classifier. GAR (genuine acceptance rate) versus FAR (false acceptance rate) will be plotted so that the performance of the biometric system can be compared using ROC (receiver operation characteristic) curve. At 85% of FAR, GAR is almost achieved 100% for all three of training data.
Palm print based biometric verification system using umace filter classification method / Lee Yong Chun