Palmprint Identification using SVM and CNN Method

Abstract

Palmprint Identification is widely used as biometrics because palms have unique and different characteristics of each person. The changes that occur in palm lines are relatively small. The data acquisition process is relatively easy and has small risks associated with the radiation effects. The palm characteristics must be processed before they can be used in biometric systems. The stage of the palmprints identification system is the enrollment and the identification stage. The preprocessing and the feature extraction method are affected to the recognition results. This paper uses a Gaussian filter for preprocessing, feature extraction using Laplacian of Gaussian and Canny edge detection, while the classification method uses Support Vector Machine and CNN. The Accuration results obtained from using Laplacian of Gaussian and Support Vector Machine are 88,3% for 60 classes with 420 images, while for CNN, an accuration rate is 97%.