Shankar Bhausaheb Nikam

Texture and Wavelet-Based Spoof Fingerprint Detection for Fingerprint Biometric Systems

This paper describes an image-based system to detect spoof fingerprint attacks in fingerprint biometric systems. It is based on the observation that, real and spoof fingerprints exhibit different textural characteristics. These are based on structural, orientation, roughness, smoothness and regularity differences of diverse regions in a fingerprint image. Local binary pattern (LBP) histograms are used to capture these textural details. Wavelet energy features characterizing ridge frequency and orientation information are also used for improving the efficiency of the proposed method. Dimensionality of the integrated feature set is reduced by running Pudilpsilas Sequential Forward Floating Selection (SFFS) algorithm. We propose to use a hybrid classifier, formed by fusing three classifiers: neural network, support vector machine and k-nearest neighbor using the ldquoProduct Rulerdquo. Classification rates achieved with these classifiers, including a hybrid classifier are in the range ~94% to ~97%. Experimental results indicate that, the new liveness detection approach is a very promising technique, as it needs only one fingerprint and no extra hardware to detect vitality.

Curvelet-based fingerprint anti-spoofing

Existing perspiration-based liveness detection algorithms need two successive images (captured in certain time interval), hence they are slow and not useful for real-time applications. Liveness detection methods using extra hardware increase the cost of the system. To alleviate these problems, we propose new curvelet-based method which needs only one fingerprint to detect liveness. Wavelets are very effective in representing objects with isolated point singularities, but fail to represent line and curve singularities. Curvelet transform allows representing singularities along curves in a more efficient way than the wavelets. Ridges oriented in different directions in a fingerprint image are curved; hence curvelets are very significant to characterize fingerprint texture. Textural measures based on curvelet energy and co-occurrence signatures are used to characterize fingerprint image. Dimensionalities of feature sets are reduced by running Pudil’s sequential forward floating selection (SFFS) algorithm. Curvelet energy and co-occurrence signatures are independently tested on three different classifiers: AdaBoost.M1, support vector machine and alternating decision tree. Finally, all the aforementioned classifiers are fused using the “Majority Voting Rule” to form an ensemble classifier. A fingerprint database consisting of 185 real, 90 Fun-Doh and 150 Gummy fingerprints is created by using varieties of artificial materials for casts and moulds of spoof fingerprints. Performance of the new liveness detection approach is found very promising, as it needs only one fingerprint and no extra hardware to detect vitality.

Normalized Cross-Correlation Based Fingerprint Matching

To perform fingerprint matching based on the number of corresponding minutia pairings, has been in use for quite sometime. But this technique is not very efficient for recognizing the low quality fingerprints. To overcome this problem, some researchers suggest the correlation technique which provides better result. Use of correlation-based methods is increasing day-by-day in the field of biometrics as it provides better results. In this paper, we propose normalized cross-correlation technique for fingerprint matching to minimize error rate as well as reduce the computational effort than the minutiae matching method. The EER (equal error rate) obtained from result till now with minutiae matching method is 3%, while that obtained for the method proposed in this paper is approx 2% for all types of fingerprints in combined form.

Wavelet-based multiresolution analysis of ridges for fingerprint liveness detection

In this paper, a new wavelet-based perspiration detection algorithm is proposed for fingerprint liveness detection. It is based on processing time-series ridge lines in the wavelet domain. The existing perspiration detection algorithm proposed in the literature captures perspiration information by processing ridge lines in the time (spatial) domain. However, for some kinds of fingers (e.g., dry and perspiration-saturated fingers), changes in perspiration are minute. These changes are difficult to extract from the grey-level intensities processed in the time domain. Due to this, such fingers may be misclassified, thus reducing overall accuracy. In practice, we often encounter poor quality, dry or wet fingers. Therefore, it is necessary to take due care of such fingers, and have an enhanced algorithm that can process these fingers as well. To alleviate the problem, this paper discusses a new algorithm that processes time-series ridge lines using the multiresolution theory of wavelets. Major sweating changes are extracted at the coarse level, and then resolution is gradually increased to notice minute details. Such a coarse-to-fine strategy provides us with rich sweating information compared to that obtained directly from grey-level intensities in the time domain, which naturally leads to improved liveness results.

Combining Gabor Local Texture Pattern and Wavelet Global Features for Fingerprint Matching

Fingerprint verification is one of the most well-known and publicized biometrics. Fingerprints have been matched over the decades using minutiae features. However, minutiae extraction from poor-quality or incomplete images is not reliable and often gives erroneous results in matching. Further, due to variations caused by the users, minutiae alignment and matching may not be exact. Fingerprints exhibit oriented texture-like patterns. Gabor filters can optimally capture local frequency and orientation information even from poor-quality or incomplete images. Like-wise, global texture information can be efficiently extracted using wavelet transform. This paper proposes a hybrid fingerprint verification system based on local texture pattern obtained using Gabor filtering and wavelet global features obtained by multiresolution analysis of a fingerprint. Experimental results show that the proposed hybrid system is efficient and suitable for real-time authentication applications with a small size database.

Level 2 features and wavelet analysis based hybrid fingerprint matcher

In this work, we present a hybrid fingerprint verification system based on Level 2 features i.e. minutiae and multiresolution analysis of fingerprint images. Systems based only on minutiae features do not perform well for poor quality images. In practice, we often encounter extremely dry, wet fingerprint images with cuts, warts, etc. Due to such fingerprints, minutiae based systems show poor performance for real time authentication applications with large number of identities. To alleviate the problem of poor quality fingerprints, and to improve overall performance of the system, we have proposed hybrid fingerprint verification based on both minutiae features and wavelet statistical features. Final matching score is calculated by fusing two matching scores of minutiae based method and wavelet based algorithm. Proposed system is tested on DB1 (set A) database of FVC 2004. The experimental results have shown that, proposed approach is more efficient and suitable than conventional minutiae based methods for real time authentication systems with large size databases.

CO-OCCURRENCE PROBABILITIES AND WAVELET-BASED SPOOF FINGERPRINT DETECTION

Perspiration phenomenon is very significant to detect the liveness of a finger. However, it requires two consecutive fingerprints to notice perspiration, and therefore may not be suitable for real ...