Lifang Wu

A novel key generation cryptosystem based on face features

With the development of Internet, Information security is becoming more and more important. Traditional cryptographic methods require the user to remember keys, it is not convenient. Biometrics based cryptographic key generation techniques generate cryptographic keys from biometrics directly. In this paper, we propose a biométrie cryptosystem based on face biometrics. At encryption stage, a 128-dimensional principal component analysis (PCA) feature vector is firstly extracted from the face image. And a 128 bit binary vector is obtained by thresholding. Then we select the distinguishable bits to form bio-key and the optimal bit order number is saved in a look-up table. Furthermore, an error-correct-code (ECC) is generated using Reed-Solomon algorithm. The message is encrypted using symmetric DES with bio-key. In decryption phase, a 128-dimensional PCA features vector extracted from the query face image. Then a bio-key is generated using the look-up table generated at encryption stage. The final key is obtained using both bio-key and Error correct code (ECC). Finally, the symmetric DES decryption algorithm implemented to obtain message using final key. The proposed scheme is tested using ORL face database, the experimental results shows that our algorithm is effective.

An Illumination Invariant Face Recognition Scheme to Combining Normalized Structural Descriptor with Single Scale Retinex

Illumination variation is still a challenging issue to address in face recognition. Retinex scheme is effective to face images under small illumination variation, but its performance drop when illumination variation is large. We further analyze the normalized images under large illumination variation and we find that the illumination variation has not been removed thoroughly in these images. Structural similarity is one of image similarity metrics similar to human perception. From SSIM, we extract the structure related component and name it as Normalized Structure Descriptor. It is clear that NSD is robust to illumination variation. We propose a scheme to combining Normalized Structural Descriptor with Single Scale Retinex. In our scheme NSD is extracted from the normalized image from SSR. And the face recognition is performed by the similarity of NSD. The experimental results on the Yale Face Database B and Extended Yale Face Database B show that our approach has performance comparable to state-of-the-art approaches.

A Face Liveness Detection Scheme to Combining Static and Dynamic Features

Face liveness detection is an interesting research topic in face-based online authentication. The current face liveness detection algorithms utilize either static or dynamic features, but not both. In fact, the dynamic and static features have different advantages in face liveness detection. In this paper, we discuss a scheme to combine dynamic and static features that combines the strength of each. First, the dynamic maps are obtained from the inter frame motion in the video. Then, using a Convolutional Neural Network (CNN), the dynamic and static features are extracted from the dynamic maps and the images, respectively. Next, the fully connected layers from the CNN that include the dynamic and static features are connected to form the fused features. Finally, the fused features are used to train a two-value Support Vector Machine (SVM) classifier, which classify the images into two groups, images with real faces and images with fake faces. We conduct experiments to assess our algorithm that includes classifying images from two public databases. Experimental results demonstrate that our algorithm outperforms current state-of-the-art face liveness detection algorithms.

Live Face Detection by Combining the Fourier Statistics and LBP

With the development of E-Commerce, biometric based on-line authentication is more competitive and is paid more attentions. It brings about one of hot issues of liveness detection recently. In this paper, we propose a liveness detection scheme to combine Fourier statistics and local binary pattern (LBP). First, The Gamma correction and DoG filtering are utilized to reduce the illumination variation and to preserve the key information of the image. Then the Fourier statistics and LBP are combined together to form a new feature vector. Finally, a SVM classifier is trained to discriminate the live and forge face image. The experimental results on the NUAA demonstrate that the proposed scheme is efficient and robust.

A Fuzzy Vault Scheme for Ordered Biometrics

The fuzzy vault scheme has recently become popular approaches to biometric template protection. Since the original scheme has been designed to work with unordered biometric features, such a scheme cannot effectively utilize order information. We present in this paper a new fuzzy vault scheme that can effectively utilize the ordered characteristics of biometric features. In this scheme, we develop ordered fuzzy vault encoding and decoding processes in order to utilize the ordered information of the features. This prevents the feature components from cross matching and reduces false acceptance ratio (FAR).xa0 Furthermore, the original biometric features (or original template) are transformed into binary features (or secure template) by random transformation. The transformed secure template provides both diversity and revocability. This transform also prevents an adversary from obtaining the original biometric template from the secure template and therefore enhance the secure level of the scheme. Based on the proposed scheme, we design an online authentication application framework implemented using face images. We compare our scheme with two contemporary approaches to verify the effectiveness of this approach. Experimental results show that our scheme is able to achieve an improved performance with several desired properties of an online authentication system.

A Secure Face-Verification Scheme Based on Homomorphic Encryption and Deep Neural Networks

With the increase in applications of face verification, increasing attention has been paid to their accuracy and security. To ensure both the accuracy and safety of these systems, this paper proposes an encrypted face-verification system. In this paper, face features are extracted using deep neural networks and then encrypted with the Paillier algorithm and saved in a data set. The framework of the whole system involves three parties: the client, data server, and verification server. The data server saves the encrypted user features and user ID, the verification server performs verification, and the client is responsible for collecting a requester’s information and sending it to the servers. The information is transmitted among parties as cipher text, which means that no parties know the private keys except for the verification server. The proposed scheme is tested with two deep convolutional neural networks architectures on the labeled faces in the Wild and Faces94 data sets. The extensive experimental results, including results for identification and verification tasks, show that our approach can enhance the security of a recognition system with little decrease in accuracy. Therefore, the proposed system is efficient with respect to both the security and high verification accuracy.

Complete Pose Binary SIFT for Face Recognition with Pose Variation

Some pose invariant face recognition approaches require preprocessing such as face alignment or landmark fitting, which is another unresolved problem. SIFT based face recognition schemes could resolve the problem of constrained pose variation without such preprocessing. we find that the sift descriptors are robust to off-plane rotation within 25 degree and in-plane rotation. Furthermore, we propose complete pose binary SIFT (CPBS) to address the issue of arbitrary pose variation. First, five face images with poses of frontal view, rotation left/right 45 and 90 degree respectively are selected as gallery images of a subject. Then the binary descriptors of these images are pooled together as CPBS of the subject. Face recognition is finished by hamming distance between the probe face image and the CPBS. Experimental results on the CMU-PIE and FERET face databases show that our approach has performance comparable to state-of-the-art approaches, while not requiring face alignment or landmark fitting.

A fuzzy vault scheme for feature fusion

Widespread application of biometric authentication brings about new problem of privacy. Biometric template protection is becoming a hot research. Efficient feature fusion is deemed to have good performance possibly. In this paper we proposed a fuzzy vault scheme for feature fusion. In our scheme, two facial features Multi-Block Local Binary Pattern (MB-LBP) and Principal Component Analysis (PCA) coefficients are extracted. A key is split into two overlapped subkeys. One is utilized to generate a set of helper data from MB-LBP. The other is utilized to generate another set of helper data from PCA coefficients. Two sets of helper data are submerged into the chaff points set and the final fuzzy vault is generated. In the fuzzy vault decoding, the MB-LBP and PCA coefficients of the query face image are utilized to recover two subkeys from the fuzzy vault. The final key is obtained from two subkeys. Because two subkeys are overlapped and complementary to each other, our scheme can obtain good authentication performance. It is confirmed by the experimental results.

Motion Analysis Based Cross-Database Voting for Face Spoofing Detection

With the rapid development of face recognition systems in various practical applications, numerous face spoofing attacks under different environment and devices have emerged. The countermeasure of face spoofing attacks in cross-database have caused increasing attention. This paper proposes a face spoofing detection method with motion analysis based cross-database voting. We employ the consistency motion information of different databases like eye-blink, mouth movements and facial expression etc. Then the motion information maps of a video is classified to real or fake by CNN model. Furthermore, cross-database voting strategy is constructed to transfer motion characteristics from a database to another for face spoofing inference. Experimental results demonstrate that the proposed method outperforms its comparisons taking benefits of motion analysis based CNN classification and cross-database voting.

An Effective Face Verification Algorithm to Fuse Complete Features in Convolutional Neural Network

Face verification for on line application is a difficult problem and many researchers have tried to solve it by convolutional neural network. Among of them, most works used the last-hidden layer as the feature of face, and abandoned the features in the lower layers which indicate local information. To remedy this, we extract features of all layers in the convolutional neural network, and fuse these features together after dimensionality reduction with PCA. Then these features are utilized for face verification with neural network classifier. Experiment results show that complete features can improve the verification rate effectively than using the last-hidden layer only.