Palmprint verification based on genetic algorithm

This paper proposes a novel idea based feature selection in the verification system of palmprint, which can realize the specific feature selection for different user using genetic algorithm (GA). In the stage of enrollment, discrete wavelet transforms (DWT) and statistical methods are first used for feature extraction. Then GA is employed for feature selection, which means that each user has a specific feature index and verification modality. In the stage of verification, according to the feature index, the selected features are input to the related support vector machine (SVM) for classification. We test the proposed method on Hong Kong Polytechnic University Palmprint Database. Through the comparison, the experimental results show that our proposed method is effective and can reach high performance.