IDENTIFIKASI CITRA SIDIKJARI MENGGUNAKAN ALIHRAGAM WAVELET DAN JARAK EUCLIDEAN

Abstract

During this time, fingerprint identification is performed manually using the naked eyes. Therefore, a research must be done to help people identifying a fingerprint. Techniques or methods which can recognize the fingerprint pattern of a program supported by the software. In this final project the system established has a purpose to recognize a fingerprint image and in the end can identify the owner of the fingerprint. This research is also to obtain a best wavelet for fingerprint recognition and to know performance from this recognition system. Fingerprint identification procces beginning from image data acquisition, image prosccesing, and decomposition procces using wavelet transform. Wavelet transform used is fingerprint image feature. Then the energy value of each fingerprint from wavelet decomposition is calculated. There,the value a energy is saved in database. The futher procces to compare the energy between fingerprint which will be identified by fingerprint which exist in database. The fingerprint energy which has a minimum Euclidean distance will be taken as the identified fingerprint. The experiment were carried out in four kind of wavelets i.e Haar, Daubiechies, Symlet and Coiflet. This experiment used 200 fingerprint image from 10 responders. Each responder has 10 database image and 10 test image. From four kind of wavelet tested, it can be shown that all of them have good result. However for best type of wavelet Daubiechies_8 has a best image recognition with recognition success rate reaches 91% without treshold value. The experiment is also done by using 5 database images and one database image. In experiment using 5 database images, recognition success rate reaches 90%, while for best wavelet experiment using 1 database image recognition success rate occurs to be declined become 83%. Keywords: Fingerprint, image proccesing, wavelet transform, Euclidean distanc