Palmprint Recognition Using PCA and Weighted Feature Level Fusion of 2D–Gabor and Log-Gabor Features

Palmprint biometric technology is most accurate and reliable, which has acquired good impact over the remaining biometric technologies. Palmprint contains various features like minutiae points, wrinkles, palm lines and texture, etc. A Number of line based approaches, subspace based approaches and texture based methods for extracting features from Palmprint have been considered and studied thoroughly. This paper presents a multi-algorithm based Palmprint recognition system. In which, from preprocessed palmprint image features are extracted by means of 2DGabor filter and Log Gabor filter. The system performance has been evaluated by considering each texture features individually. Then, these texture features are combined using weighted feature level fusion. As fusion at feature level leads to high dimensional data, Principal Component Analysis (PCA) has been applied to reduce the dimension. Based on Euclidean distance the palmprints have been matched. The experiments have been done on IIT Delhi database. And the results had shown that the proposed system has significant improvement in accuracy when compared against individual recognition systems.