IDENTIFIKASI TELINGA BERDASARKAN FITUR GEOMETRIK; EAR IDENTIFICATION USING GEOMETRIC FEATURE

Siregar, Alda Cendekia, Agus Harjoko

2015 | Disertasi | FMIPA

Nowadays, needs for identification system is increasing. Biometric is considered as one of the most robust human identification method due to its high level of security and accuracy. Ear has been introduced as biometric recently. Ear has distinct feature that possess unique value for every individuals. One of ear features which can be used to differentiate one person from another is geometrical feature. Feature extraction produces the ratio of long axis to each feature point. This ratio are invariant to rotation, scaling and translation. System performance is evaluated using accuracy measure. Result of this research shows that RBFNN has 82,67% accuracy. Accuracy comparison is conducted among other classifier i.e. KNN and MLP backpropagation. KNN has 97% accuracy and MLP backropagation has 79,67% accuracy. RBFNN’s accuracy higher then MLP’s.

Kata Kunci : Ear identification; geometrical feature; RBFNN; KNN; MLP Backpropagation.