Hand-Based Personal Identification Using K-Means Clustering and Modified Zernike Moments

Biometrics-based identity verification is regarded as an effective approach for automatic recognition recently. A novel personal identity verification approach based-on palmprint is proposed in this paper. Both a coarse-to-fine identification strategy and the weight-based self-adaptive feature selection mechanism are adopted to facilitate the verification task and improve veracity. The wavelet transformation and modified Zernike moments techniques are used to extract the texture features of palmprint. In the identification stage, the K-means algorithm is first used to select a small set of similar candidates from the database for further matching. After feature optimization, one-class-one-network (back propagation neural network (BPNN)) classification structure is employed for final determination. The experimental results show the proposed methods are effectiveness and accuracy.