Fingerprint Liveness Detection From Single Image Using Low-Level Features and Shape Analysis

ABSTRACT:

Fingerprint-based authentication systems have developed rapidly in the recent years. However, current fingerprint-based biometric systems are vulnerable to spoofing attacks. Moreover, single feature-based static approach does not perform equally over different fingerprint sensors and spoofing materials. In this paper, we propose a static software approach. We propose to combine low-level gradient features from speeded-up robust features, pyramid extension of the histograms of oriented gradient and texture features from Gabor wavelet using dynamic score level integration. We extract these features from a single fingerprint image to overcome the issues faced in dynamic software approaches, which require user cooperation and longer computational time. A experimental analysis done on LivDet 2011 data produced an average equal error rate (EER) of 3.95% over four databases. The result outperforms the existing best average EER of 9.625%. We also performed experiments with LivDet 2013 database and achieved an average classification error rate of 2.27% in comparison with 12.87% obtained by the LivDet 2013 competition winner.

PROJECT OUTPUT VIDEO:

EXISTING SYSTEM:

• In existing system static software based approach, a user is only required to place his/her finger on the sensor for a short duration in an undedicated way for a single image capture. Most of the works in fingerprint liveness detection use a single feature based approach.

• For example, the works in previous method engineer’s the feature extracted from a specific material for detecting fake fingerprint.

DISADVANTAGES OF EXISTING SYSTEM:

• Based on our analysis, we determine that a single feature set from a single classifier is insufficient to perform similarly over different databases which are recorded using different fingerprint sensors. This is because different sensors capture information differently.

• In addition, various materials such as gelatin based fake fingerprint may not produce similar features as compared to other materials such as latex, play-doh or wood-glue. This is because fake fingerprints exhibit different intensity gradient and ridge shape due to the thickness of material used.

• The process of creating fake fingerprint also introduces air bubbles. Furthermore, it is not practical for the authentication system to have prior knowledge of the types of material used to create the fake fingerprint in real world scenarios.

PROPOSED SYSTEM:

• In this paper, we propose a method to overcome the limitations faced in the static software based approaches where a single feature set fails to perform equally over different fingerprint sensors and materials.

• Our methodology extracts low level textural and gradient information for fingerprint liveness detection from a single image. We propose the use of SURF features in combination with PHOG to obtain gradient features that discriminate well between fake and live fingerprint images. SURF features have a concise descriptor length which is compact and takes less computational time as compared to LBP. In addition, SURF is also invariant to scale and image rotation.

• PHOG feature descriptor extracts intensity gradient and edge directions to describe the shape and appearance in an image. The PHOG extractor is also invariant to geometric and photometric transformation. Thus, combination of SURF and PHOG enables our method to perform similarly over various sensors and materials.

• In order to obtain textural features, we propose the use of Gabor wavelets as they have optimal localization properties in both the frequency and spatial domain. They extract discriminative ridge feature maps and have performed well in discriminating between live and fake fingerprint images.

ADVANTAGES OF PROPOSED SYSTEM:

• To the best of our knowledge, the proposed method is one of the few works that performs well over a large open source dataset created using six different sensors and six different materials. In our paper, we investigate the use of local discriminative feature space on live and spoof fingerprints by using PHOG, SURF, GABOR and their combinations.

• Experiments performed on six sensors demonstrate that the combination of PHOG and SURF always works better than PHOG and SURF individually for LivDet 2011 and 2013 databases. This indicates that these descriptors complement each other. Also, the combination of PHOG and SURF feature vector produces a strong discriminative feature vector which performs remarkably well in the field of fingerprint liveness detection.

• Unlike, LivDet 2013 competition winner and other top four teams which do not perform well on Crossmatch sensor, our method performs exceedingly well on Crossmatch sensor producing an average classification error of 2.5% compared to 31.20% achieved in LivDet 2013 fingerprint competition.

• The method we proposed is purely software based and it is computationally cheap, fast and flexible for future adaptations. This method can be deployed in real-time applications. Finally, the result achieved by our method outperforms the state of the art significantly.

MODULES:

• Image Acquisition

• Preprocessing

• Feature Extraction

• Feature Selection

• Classification

MODULE DESCRIPTION:

Image Acquisition:

• Image acquisition in image processing can be broadly defined as the action of retrieving an image from some source, usually a hardware-based source, so it can be passed through whatever processes need to occur afterward.

• Performing image acquisition in image processing is always the first step in the workflow sequence because, without an image, no processing is possible. The image that is acquired is completely unprocessed and is the result of whatever hardware was used to generate it, which can be very important in some fields to have a consistent baseline from which to work.

Preprocessing:

• The aim of pre-processing is an improvement of the image data that suppresses unwanted distortions or enhances some image features important for further processing.

• We enhanced the quality of the image by first cropping the fingerprint region in the image and median filtering is then applied on the cropped images without reducing the sharpness of the input image. Finally, histogram equalization is performed to improve the contrast in the image by diversifying the intensity range over the whole cropped image. The output achieved after this stage is an image with a reduced noise and improved definition of the ridge structure.

Feature Extraction:

• In fingerprint authentication systems, the image is usually captured from multiple subjects using different scanners. Therefore, fingerprint images are typically found to be of different scales and rotations. In certain scenarios, the fingerprint images are partially captured due to human errors. In order to obtain features that are invariant to these problems, we use various features that capture properties of live fingerprint images.

• In our work, we choose to use SURF as it is invariant to illumination, scale and rotation. SURF is also used because of its concise descriptor length. While SURF is invariant to object orientation and scale transformation, it is not invariant to geometric transformations. Hence, in order to compensate the limitations of SURF, PHOG descriptors are used to extract local shape information to obtain more discriminative features. In addition, Gabor wavelet features are also incorporated for texture analysis.

Feature Reduction using PCA:

• Excessive features increase computation times and storage memory. Furthermore, they sometimes make classification more complicated, which is called the curse of dimensionality. It is required to reduce the number of features. PCA is an efficient tool to reduce the dimension of a data set consisting of a large number of interrelated variables while retaining most of the variations. It is achieved by transforming the data set to a new set of ordered variables according to their variances or importance.

Classification:

• The classification process is done over the extracted features. The main novelty here is the adoption of SVM and Random Forest. RF and SVM classifier is applied over the features and the classification is done.

SYSTEM REQUIREMENTS:

HARDWARE REQUIREMENTS:

• System            : Pentium Dual Core.

• Hard Disk       : 120 GB.

• Monitor          : 15’’ LED

• Input Devices : Keyboard, Mouse

• Ram                : 1GB.

SOFTWARE REQUIREMENTS:

• Operating system : Windows 7.

• Coding Language : MATLAB

• Tool                       : MATLAB R2013A

REFERENCE:

Rohit Kumar Dubey, Jonathan Goh, and Vrizlynn L. L. Thing, “Fingerprint Liveness Detection From Single Image Using Low-Level Features and Shape Analysis”, IEEE TRANSACTIONS ON INFORMATION FORENSICS AND SECURITY, VOL. 11, NO. 7, JULY 2016.