A novel Fingervein Recognition System based on Monogenic Local Binary Pattern Features

As a new approach to human identification, fingervein recognition is becoming an active biometric recognition mode. This paper focuses on fingervein recognition system. First, a preprocessing algorithm is used to enhance each fingervein image. Then, an improvement technique of feature extraction based on Monogenic Local Binary Pattern (MLBP) is presented. This novel metric integrates the conventional LBP (Local Binary Pattern) with the other two rotation invariant measures (local phase and local surface type) to lower the computational complexity while slightly increasing the matching accuracy. Experimental results show that the proposed algorithm offres best performances in fingervein recognition. In fact, the area under curve of proposed approach has very close to unity (0.91) Index Terms keywordFingervein, texture, classification, feature extraction, local phase, local surface type, MLBP.