Fake fingerprint liveness detection based on micro and macro features

The most established technique for the identification at biometric access control systems is the human fingerprint. While every human fingerprint is unique, fingerprints can be faked very easily by using thin layer fakes. Because commercial fingerprint scanners use only a two-dimensional image acquisition of the finger surface, they can only hardly differentiate between real fingerprints and fingerprint fakes applied on thin layer materials. A Swept Source OCT system with an A-line rate of 20 kHz and a lateral and axial resolution of approximately 13 μm, a centre wavelength of 1320 nm and a band width of 120 nm (FWHM) was used to acquire fingerprints and finger tips with overlying fakes. Three-dimensional volume stacks with dimensions of 4.5 mm x 4 mm x 2 mm were acquired. The layering arrangement of the imaged finger tips and faked finger tips was analyzed and subsequently classified into real and faked fingerprints. Additionally, sweat gland ducts were detected and consulted for the classification. The manual classification between real fingerprints and faked fingerprints results in almost 100 % correctness. The outer as well as the internal fingerprint can be recognized in all real human fingers, whereby this was not possible in the image stacks of the faked fingerprints. Furthermore, in all image stacks of real human fingers the sweat gland ducts were detected. The number of sweat gland ducts differs between the test persons. The typical helix shape of the ducts was observed. In contrast, in images of faked fingerprints we observe abnormal layer arrangements and no sweat gland ducts connecting the papillae of the outer fingerprint and the internal fingerprint. We demonstrated that OCT is a very useful tool to enhance the performance of biometric control systems concerning attacks by thin layer fingerprint fakes.

Fingerprint is widely used physical human trait for uniquely identification and verification of human but use of fingerprint is becoming very challenging due to spoofing attacks. In this paper, we present new combination of local diagonal extrema pattern and local phase quantization descriptors to extract the features to generate feature vector for training and testing fingerprint images. Local diagonal extrema pattern finds the relationship between center pixels with diagonal neighboring pixels and local phase quantization extracts the local phase information by using short-term Fourier transform. LDEP reduces the dimensionality problem and forms a good combination of feature descriptor with LPQ. Combined extracted features of training and testing images using both descriptors are passed to Support Vector Machine for discriminating live and fake fingerprints. Experiments have been performed on LivDet2009 dataset and results show the effective and efficient performance of the proposed system. The proposed system achieved good accuracy and very less error rate in comparison to the different descriptors.

In recent decades, the classification of fake fingerprints and real fingerprint images has become an attractive research topic because of the most advanced threats. A number of research works have been carried out to classify fake and real fingerprints. But, most of the existing techniques did not utilize swarm intelligence techniques in their fingerprint classification system. Swarm intelligence has been widely used in various applications due to its robustness and potential in solving a complex optimization problem. The proposed classification methodology comprises of four steps, namely image preprocessing, feature extraction, feature selection and classification. This work uses efficient min-max normalization and median filtering for preprocessing, and multiple static features are extracted from Gabor filtering. To perform multiple static feature selection, Artificial Bee Colony (ABC) and Modified Artificial Bee Colony (MABC) optimization algorithm is used in this work to choose the best optimal static features on the basis of specific fitness values. This approach uses Fuzzy Feed Forward Neural Network (FFFNN) for classification and classification of feature vector into fake and real fingerprint images by the semi supervised graph based classification approach which enables efficient classification of real and fake fingerprints and its performance is compared with the SVM classifier. Evaluation of the performance has been carried out by taking the fingerprint images that had been collected from the FVC2000 and synthetically generated database by employing the SFinGE. The results showed that the proposed work provided improvised and enhanced results with respective to characteristics like sensitivity, precision, specification as well as classification accuracy.

With the increasing use of fingerprint authentication systems on portable devices, fake fingerprint detection has become growing important because fingerprints can be easily spoofed from a variety of available fabrication materials. Recently, many smartphones are hacked successfully by 2D fake fingerprint, which is a serious threat to authentication security. In order to enhance the robustness against fake fingerprint type, this paper proposes a novel 2D fake fingerprint detection method for portable devices based on improved light Convolutional Neural Networks (CNN). To evaluate the performance of the proposed method, a new 2D fake fingerprint dataset including three fabrication materials is created from capacitive fingerprint scanner. In addition, batch normalization and global average pooling are integrated to optimize the network. Experimental results show that the proposed method has high accuracy, strong robustness and good real-time performance, and can meet the requirements on portable devices.

Fingerprint recognition is very stable, reliable, and oldest biometric authentication technology that has been widely used in various civilian and government applications. Like other authentication technologies, fingerprint recognition is not totally free from spoofing. A fingerprint recognition system can be spoofed by placing artificially generated fingerprints of enrolled user at the sensor that are made up of materials like gelatine, silicon, wax or play-doh etc. Liveness detection is a technique to enhance the security of such systems by detecting physiological life signs in fingerprint samples to differentiate a real fingerprint from an artificial one. A live fingerprint image has the property to produce gray level variations along ridges due to presence of sweat pores, perspiration and skin quality (dry, wet) whereas a fake fingerprint has uniform gray levels that can give a fair idea about the liveness in input sample. In the proposed work, two fingerprints are captured by scanner at different time points (0 sec and 2 sec). In order to find their gray level variations, a gray level co-occurrence matrix of each fingerprint image is determined. After that, contrast value is computed from each matrix. If the difference in their contrast value is greater than threshold, then the input fingerprint sample is considered as live otherwise it is considered as fake. Liveness detection solution presented in this work helps to avoid direct attacks at fingerprint sensor that result in very secure and precise fingerprint detection. Keywords: Biometrics, Fingerprint detection, Liveness Detection, Spoofing

We address the problem of localizing objects using color, texture and shape. Given a handrawn sketch for querying an object shape, and its color and texture, the algorithm automatically searches the database images for objects which meet the query attributes. The database images do not need to be presegmented or annotated. The proposed algorithm operates in two stages. In the first stage, we use local texture and color features to find a small number of candidate images, and identify regions in the candidate images which share similar texture and color as the query example. To speed up the processing, the texture and color features are directly extracted from the Discrete Cosine Transform (DCT) compressed domain. In the second stage, we use a deformable template matching method to match the query shape to the image edges at the locations which possess the desired texture and color attributes. This algorithm is different from the other content-based image retrieval algorithms in that: (i) no presegmentation of the database images is needed, and (ii) the color and texture features are directly extracted from the compressed images. Experimental results show that substantial computational savings can be achieved utilizing multiple image cues.

Object localization using color, texture and shape

UEFSD: Unsupervised medical images fusion based on exclusive features and saliency detection for SPECT-MRI images of glioma

Biometric authentication can establish a person’s identity from their exclusive features. In general, biometric authentication can vulnerable to spoofing attacks. Spoofing referred to presentation attack to mislead the biometric sensor. An anti-spoofing method is able to automatically differentiate between real biometric traits presented to the sensor and synthetically produced artifacts containing a biometric trait. There is a great need for a software-based liveness detection method that can classify the fake and real biometric traits. In this paper, we have proposed a liveness detection method using fingerprint and iris. In this method, statistical texture features and spatial analysis of the fingerprint pattern is utilized for fake or real classification. The approach is further improved by fusing iris modality with the fingerprint modality. The standard Haralick’s statistical features based on the gray level co-occurrence matrix (GLCM) and Neighborhood Gray-Tone Difference Matrix (NGTDM) are used to generate a feature vector from the fingerprint. Texture feature from iris is used to boost the performance of the proposed liveness detection method. For the fusion Dempster-Shafer (D-S) approach is used at the decision level. Experiments have been performed on ATVS dataset and LivDet2011 dataset. The results show the convincing and effective outcomes of the proposed method.

We investigated the application of multiresolution global and local texture features to reduce false-positive detection in a computerized mass detection program. One hundred and sixty-eight digitized mammograms were randomly and equally divided into training and test groups. From these mammograms, two datasets were formed. The first dataset (manual) contained four regions of interest (ROIs) selected manually from each of the mammograms. One of the four ROIs contained a biopsy-proven mass and the other three contained normal parenchyma, including dense, mixed dense/fatty, and fatty tissues. The second dataset (hybrid) contained the manually extracted mass ROIs, along with normal tissue ROIs extracted by an automated Density-Weighted Contrast Enhancement (DWCE) algorithm as false-positive detections. A wavelet transform was used to decompose an ROI into several scales. Global texture features were derived from the low-pass coefficients in the wavelet transformed images. Local texture features were calculated from the suspicious object and the peripheral subregions. Linear discriminant models using effective features selected from the global, local, or combined feature spaces were established to maximize the separation between masses and normal tissue. Receiver Operating Characteristic (ROC) analysis was conducted to evaluate the classifier performance. The classification accuracy using global features were comparable to that using local features. With both global and local features, the average area, Az, under the test ROC curve, reached 0.92 for the manual dataset and 0.96 for the hybrid dataset, demonstrating statistically significant improvement over those obtained with global or local features alone. The results indicated the effectiveness of the combined global and local features in the classification of masses and normal tissue for false-positive reduction.

In content based image retrieval (CBIR) process, every image has been represented in a compact set of local visual features i.e. color, texture, and/or shape of images. This set of local visual features is known as feature vector. In the CBIR process, feature vectors of images have been used to represent or to identify similar images in adequate way. As a result, feature vector construction has always been considered as an important issue since it must reflect proper image semantics using minimal amount of data. The proposed CBIR scheme is based on the combination of color and texture features. In this work initially, we have converted the given RGB image into HSV color image. Subsequently, we have considered H (hue), S (saturation), and V (intensity) components for extraction of visual image features. The texture features have been extracted from the V component of the image using 2D dual-tree complex wavelet transform (2D DT-CWT) where it analyzes the textural patterns in six different directions i.e. ±15\(^{\circ }\), ±45\(^{\circ }\), and ±75\(^{\circ }\). At the same time, we have computed the probability histograms of H and S components of the image respectively and subsequently those are divided into non-uniform bins based on cumulative probability for extraction of color based features. So, in this work both the color and texture features have been extracted simultaneously. Finally, the obtained features have been concatenated to attain the final feature vector and same is considered in image retrieval process. We have tested the novelty and performance of the proposed work in two Corel, two objects, and, a texture image datasets. The experimental results reveal the acceptable retrieval performances for different types of datasets.

Comparative Analysis of Different Texture Features in Breast Abnormality Prediction

Batik is a traditional craft of designing patterned fabrics which hold high artistic value in Sri Lankan culture, where hand-painted wax patterns are coloured using specialist dyeing methods to create the finished product. This paper presents a study of vision-based feature extraction of Batik images considering colour, texture and shape features to develop a comprehensive feature descriptor of Batik motifs. Wax drawn patterns are identified from the digital images of Batik motifs to retrieve an outline of patterns demarcating the different coloured layers generated by multiple stages of dyeing. Motifs with repetitive patterns are identified using the Local Binary Pattern (LBP) as a texture feature vector. Both RGB and L*a*b* colour schemes are studied in the representation of Batik motifs. The colour description is presented using Mini Batch K-Means which out-performed the widely used K-Means clustering method. Hu Moment Invariants are used for shape feature extraction, and Gray Level Co-occurrence Matrix (GLCM) for texture feature extraction. A comprehensive feature descriptor is developed to represent Batik designs, which could be used to recommend similar designs based on the shape and texture features of query images presented by the user.

Historically, behavior analytic research and practice have been grounded in the single subject examination of behavior change. Within the behavior analytic community, there remains little doubt that the graphing of behavior is a powerful strategy for demonstrating functional control by the independent variable; however, during the past thirty years, various statistical techniques have become a popular alternative form of evidence for demonstrating a treatment effect. Concurrently, a mounting number of behavior analytic investigators are measuring multiple dependent variables when conducting statistical analyses. Without employing strategies that protect the experimentwise error rate, evaluation of multiple dependent variables within a single experiment is likely to inflate the Type I error rate. In fact, with each additional dependent variable examined in univariate fashion, the probability of “incorrectly” identifying statistical significance increases exponentially as a function of chance. Multivariate analysis of variance (MANOVA) and several other statistical techniques can preclude this common error. We provide an overview of the procedural complications arising from methodologies that might inflate the Type I error rate. Additionally, we provide a sample of reviewer comments and suggestions, and an enrichment section focusing on this somewhat contentious issue, as well as a number of statistical and neural network techniques that enhance power and preclude the inflation of Type I error rates

The concept of the local climate zone (LCZ) has been recently proposed as a generic land-cover/land-use classification scheme. It divides urban regions into 17 categories based on compositions of man-made structures and natural landscapes. Although it was originally designed for temperature study, the morphological structure concealed in LCZs also reflects economic status and population distribution. To this end, global LCZ classification is of great value for worldwide studies on economy and population. Conventional classification approaches are usually successful for an individual city using optical remote sensing data. This paper, however, attempts for the first time to produce global LCZ classification maps using polarimetric synthetic aperture radar (PolSAR) data. Specifically, we first produce polarimetric features, local statistical features, texture features, and morphological features and compare them, with respect to their classification performance. Here, an ensemble classifier is investigated, which is trained and tested on already separated transcontinental cities. Considering the challenging global scope this work handles, we conclude the classification accuracy is not yet satisfactory. However, Sentinel-1 dual-Pol SAR data could contribute the classification for several LCZ classes. According to our feature studies, the combination of local statistical features and morphological features yields the best classification results with 61.8% overall accuracy (OA), which is 3% higher than the OA produced by the second best features combination. The 3% is considerably large for a global scale. Based on our feature importance analysis, features related to VH polarized data contributed the most to the eventual classification result.