Title: Feature fusion using Gabor filters and cooccurrence probabilities for fingerprint antispoofing
Authors: Shankar Bhausaheb Nikam, Suneeta Agarwal
Addresses: Department of Computer Science and Engineering, Motilal Nehru National Institute of Technology, Allahabad-211004, India. ' Department of Computer Science and Engineering, Motilal Nehru National Institute of Technology, Allahabad-211004, India
Abstract: Perspiration-based liveness detection method is slow, as it requires two consecutive fingerprints to notice perspiration. Some other methods in the literature need extra hardware to detect liveness. To alleviate the problems, a single-image method using fusion of Gabor features and grey level cooccurrence probability (GLCP) features is proposed. It is based on the observation that, real and spoof fingerprints exhibit different textural characteristics. Dimensionality of the features is reduced by principal component analysis (PCA). We test feature sets on three classifiers: AdaBoost.M1, support vector machine and alternating decision tree; then we fuse all the classifiers using the |Max Rule| to form an ensemble classifier. Fused feature set is found to produce higher accuracy (∼98.35% classification rate) relative to the individual feature sets (classification accuracy ranges from ∼93.88% to ∼96.71%). Thus, the performance of new liveness detection approach is very promising, as it needs only one fingerprint and no extra hardware to detect vitality.
Keywords: biometrics; fakes; fingerprints; Gabor filter; grey level cooccurrence probability; fingerprint anti-spoofing; spoofing; texture; feature fusion; perspiration; liveness detection; principal component analysis; PCA; classification.
International Journal of Intelligent Systems Technologies and Applications, 2009 Vol.7 No.3, pp.296 - 315
Available online: 15 Jul 2009 *Full-text access for editors Access for subscribers Purchase this article Comment on this article