A robust palmprint identification system using Histogram of Oriented Gradients and multi-classifiers

Nowadays, identification of persons has a great importance for information protection and access control. Thus, automatic person identification based on biometrics has become a focus of interest both for research and commercial purposes. Among the biometrics used, palmprint identification is one of the most stable and reliable technology. Some desirable properties such as uniqueness, stability, and non invasiveness make this technology suitable for highly reliable person identification. In this paper, a method is proposed based on Histogram of Oriented Gradients (HOG) descriptors for palmprint identification. This method utilized the fusion, at matching score level, of some classifiers (Radial Basis Function (RBF), Random Forest Transform (RTF) and Support Vector Machine (SVM)) to improve the performance in identification accuracy. Extensive experiments show the effectiveness of the proposed method with respect to the identification rate.