Abstract
Researches on fingerprint classification are generally based on its features such as core and delta. Extraction of these features are generally preceded by a variety of preprocessing. In this study the classification is done directly on the fingerprint image without preprocessing. Feature used as the basis for classification is the direction of the ridge. The direction of the ridge is determined by the slope of the blocks that are exist on every ridge. Fingerprint image is divided into blocks of size 3x3 pixels and the direction of each block is determined. Direction of the slope of the block are grouped into 8, these are north, north-east, east, south-east, south, south-west, west and north-west. The number of blocks in each direction form the basis of classification using Learning Vector Quantization network (LVQ). This study used 80 data samples from the database of FVC2004. This model obtained classification accuracy of up to 86.3%.
Keywords: fingerprint, classification, ridge, LVQ
Pendahuluan
Sidik jari sebagai alat identitas pribadi sudah lama dikenal dan akhir-akhir ini pemakaiaannya semakin meluas. Mesin presensi, Surat Ijin Mengemudi (SIM) dan Kartu Tanda Penduduk Elektronik (e-KTP) sudah memanfaatkan sidik jari. Sidik jari diyakini memiliki sifat unik dan permanen sehingga secara hukum diakui sebagai alat bukti yang sah. Salah satu persoalan dalam pencarian berkas sidik jari dalam database adalah besarnya jumlah data. Ketika jumlah sidik jari yang disimpan semakin bertambah maka persoalan tersebut semakin terasa. Untuk mengatasi persoalan tersebut dilakukan klasifikasi pada database sidik jari.
Peneliti: S. Suwarno dan A.Harjoko
Untuk lebih lengkapnya silahkan download di link berikut:
Post a Comment
Post a Comment