Unified Generative Adversarial Networks for Multidomain Fingerprint Presentation Attack Detection

With the rapid growth of fingerprint-based biometric systems, it is essential to ensure the security and reliability of the deployed algorithms. Indeed, the security vulnerability of these systems has been widely recognized. Thus, it is critical to enhance the generalization ability of fingerprint presentation attack detection (PAD) cross-sensor and cross-material settings. In this work, we propose a novel solution for addressing the case of a single source domain (sensor) with large labeled real/fake fingerprint images and multiple target domains (sensors) with only few real images obtained from different sensors. Our aim is to build a model that leverages the limited sample issues in all target domains by transferring knowledge from the source domain. To this end, we train a unified generative adversarial network (UGAN) for multidomain conversion to learn several mappings between all domains. This allows us to generate additional synthetic images for the target domains from the source domain to reduce the distribution shift between fingerprint representations. Then, we train a scale compound network (EfficientNetV2) coupled with multiple head classifiers (one classifier for each domain) using the source domain and the translated images. The outputs of these classifiers are then aggregated using an additional fusion layer with learnable weights. In the experiments, we validate the proposed methodology on the public LivDet2015 dataset. The experimental results show that the proposed method improves the average classification accuracy over twelve classification scenarios from 67.80 to 80.44% after adaptation.     

A comparison of criterion functions for fusion of multi-focus noisy images

In many practical applications, images are distorted by impulsive noise (IN) produced by image sensors and/or communication channels. This noise may cause miscalculation of sharpness values which, in turn, introduce significant errors in the results of image fusion. In this paper, conventional focus measures and frequency selective weighted median filter (FSWM) are evaluated for fusion of multi-focus images in the presence of IN. FSWM is also compared with other multi-focus fusion methods such as Laplacian Pyramid and wavelet. Experimental results are presented for several sets of images and the results show that FSWM can provide better performance than other focus measures and methods.     

Performance analysis of the virtual sound barrier system with a diffracting sphere

Since a practical virtual sound barrier (VSB) system has to be implemented surrounding a human’s head, the effects of such a diffracting sphere on the performance of the system are considered in this paper. It is found that the performance of the VSB system with a diffracting sphere might be improved or decreased depending on the size of the zone surrounded by the error sensors and the noise frequency, and the noise reduction distribution around the sphere tends to be uniform due to the diffraction of the sphere. Numerical simulations show that the tendency of the control performance with respect to the configuration of the system with a diffracting sphere is similar to that without the sphere. The diffracting sphere might move in the quiet zone of the VSB system, and the performance decreases with the increase of the distance between the sphere and the center of the system. However, the noise reduction level can remain more than 10 dB even when the sphere is near the border of the quiet zone. Experiments with a 16-channel cylindrical VSB system in a normal room validate the above conclusions.     

A unified classifier for robust face recognition based on combining multiple subspace algorithms

Face recognition being the fastest growing biometric technology has expanded manifold in the last few years. Various new algorithms and commercial systems have been proposed and developed. However, none of the proposed or developed algorithm is a complete solution because it may work very well on one set of images with say illumination changes but may not work properly on another set of image variations like expression variations. This study is motivated by the fact that any single classifier cannot claim to show generally better performance against all facial image variations. To overcome this shortcoming and achieve generality, combining several classifiers using various strategies has been studied extensively also incorporating the question of suitability of any classifier for this task. The study is based on the outcome of a comprehensive comparative analysis conducted on a combination of six subspace extraction algorithms and four distance metrics on three facial databases. The analysis leads to the selection of the most suitable classifiers which performs better on one task or the other. These classifiers are then combined together onto an ensemble classifier by two different strategies of weighted sum and re-ranking. The results of the ensemble classifier show that these strategies can be effectively used to construct a single classifier that can successfully handle varying facial image conditions of illumination, aging and facial expressions.     

Evaluation of infrared collimators for testing thermal imaging systems

Infrared reflective collimators are important components of expensive sophisticated test systems used for testing thermal imagers. Too low quality collimators can become a source of significant measurement errors and collimators of too high quality can unnecessarily increase cost of a test system. In such a situation it is important for test system users to know proper requirements on the collimator and to be able to verify its performance. A method for evaluation of infrared reflective collimators used in test systems for testing thermal imagers is presented in this paper. The method requires only easily available optical equipment and can be used not only by collimator manufactures but also by users of test equipment to verify performance of the collimators used for testing thermal imagers.     

语音情感识别中的特征选择方法

为了解决传统卷积神经网络在识别中文语音时预测错误率较高、泛化性能弱的问题,首先以深度卷积神经网络(DCNN)-连接时序分类(CTC)为研究对象,深入分析了不同卷积层、池化层以及全连接层的组合对其性能的影响;其次,在上述模型的基础上,提出了多路卷积神经网络(MCNN)-连接时序分类(CTC),并联合SENet提出了深度SE-MCNN-CTC声学模型,该模型融合了MCNN与SENet的优势,既能加强卷积神经网络的深层信息的传递、避免梯度问题,又可以对提取的特征图进行自适应重标定。最终实验结果表明:SE-MCNN-CTC相较于DCNN-CTC错误率相对降低13.51%,模型最终的错误率达22.21%;算法改进后的声学模型可以有效地提升泛化性能。     

Surface plasmon resonance characterization of thermally evaporated thin gold films

The heterogeneous character of thin gold films prepared by thermal evaporation and the dependence of this heterogeneity on the rate of their deposition must be considered when exploiting their optical properties for biosensor purposes. For instance, the performance of thin gold films for surface plasmon resonance (SPR) biosensors may drastically be degraded if care is not taken to prepare a film with a high fraction of gold (>95%). We use three different models to interpret the SPR response of gold films prepared by thermal evaporation. We show that the interpretation of the SPR curves requires considering both a global heterogeneity of the gold films and a surface roughness. Our conclusions are further corroborated by scanning surface plasmon microscope (SSPM) images of these thin gold films.     

Fisher’s linear discriminant ratio based threshold for moving human detection in thermal video

In video surveillance, the moving human detection in thermal video is a critical phase that filters out redundant information to extract relevant information. The moving object detection is applied on thermal video because it penetrate challenging problems such as dynamic issues of background and illumination variation. In this work, we have proposed a new background subtraction method using Fisher’s linear discriminant ratio based threshold. This threshold is investigated automatically during run-time for each pixel of every sequential frame. Automatically means to avoid the involvement of external source such as programmer or user for threshold selection. This threshold provides better pixel classification at run-time. This method handles problems generated due to multiple behavior of background more accurately using Fisher’s ratio. It maximizes the separation between object pixel and the background pixel. To check the efficacy, the performance of this work is observed in terms of various parameters depicted in analysis. The experimental results and their analysis demonstrated better performance of proposed method against considered peer methods.     

基于单通道算法的HJ-1B与Landsat 5 TM地表温度反演一致性研究

针对传感器参数与Landsat 5 TM较为相近的环境减灾卫星HJ-1B的热红外数据是否能达到与TM6反演地表温度同样的效果的问题,作者获取了2009年4月15日覆盖北京地区相同范围的HJ-1B影像和Landsat 5 TM影像,采用Ji m啨nez-Mu oz和Sobrino的普适性单通道算法,通过遥感数据获取反演参数,对研究区的地表温度进行反演,并对两类数据反演结果的一致性进行了分析和评价。通过对反演结果的统计分析和分级填图表明,基于普适性单通道算法的HJ-1B与TM的热红外波段反演的地表温度在数值上和空间分布趋势上一致性较高。因此,具有较高时间分辨率和较大幅宽等优势的HJ-1B IRS数据在地表温度反演方面具有更大的潜力和应用范围。     

Improved migration models of biogeography-based optimization for sonar dataset classification by using neural network

Classification of experimental datasets such as target and clutter in sonar applications is a complex and challenging problem. One of the most useful instrument to classify sonar datasets is Multi-Layer Perceptron Neural Network (MLP NN). In this paper, due to the optimally updating the weights and biases vector of the MLP NN, Biogeography-Based Optimization (BBO) is used to train the network. BBO has a fair ability to solve high-dimensional real-world problems (such as sonar dataset classification) by maintaining a suitable balance between exploration and exploitation phases. The performance of BBO is sensitive to the migration model, especially for high-dimensional problems. To improve the exploitation ability of BBO and to record the better results for classifying sonar dataset, we propose novel migration models such as exponential-logarithmic, and some improved migration models having different emigration and immigration mathematical functions. To validate the performance of the proposed classifiers, this network will classify three datasets with various sizes and complexities. The simulation results indicate that our newly proposed classifiers perform better than the other benchmark algorithms in addition to original BBO in terms of avoiding gets stuck in local minima, classification accuracy, and convergence speed.     

Approaching Adversarial Example Classification with Chaos Theory

Adversarial examples are one of the most intriguing topics in modern deep learning. Imperceptible perturbations to the input can fool robust models. In relation to this problem, attack and defense methods are being developed almost on a daily basis. In parallel, efforts are being made to simply pointing out when an input image is an adversarial example. This can help prevent potential issues, as the failure cases are easily recognizable by humans. The proposal in this work is to study how chaos theory methods can help distinguish adversarial examples from regular images. Our work is based on the assumption that deep networks behave as chaotic systems, and adversarial examples are the main manifestation of it (in the sense that a slight input variation produces a totally different output). In our experiments, we show that the Lyapunov exponents (an established measure of chaoticity), which have been recently proposed for classification of adversarial examples, are not robust to image processing transformations that alter image entropy. Furthermore, we show that entropy can complement Lyapunov exponents in such a way that the discriminating power is significantly enhanced. The proposed method achieves 65% to 100% accuracy detecting adversarials with a wide range of attacks (for example: CW, PGD, Spatial, HopSkip) for the MNIST dataset, with similar results when entropy-changing image processing methods (such as Equalization, Speckle and Gaussian noise) are applied. This is also corroborated with two other datasets, Fashion-MNIST and CIFAR 19. These results indicate that classifiers can enhance their robustness against the adversarial phenomenon, being applied in a wide variety of conditions that potentially matches real world cases and also other threatening scenarios.     

Dual-Domain Fusion Convolutional Neural Network for Contrast Enhancement Forensics

Contrast enhancement forensics techniques have always been of great interest for the image forensics community, as they can be an effective tool for recovering image history and identifying tampered images. Although several contrast enhancement forensic algorithms have been proposed, their accuracy and robustness against some kinds of processing are still unsatisfactory. In order to attenuate such deficiency, in this paper, we propose a new framework based on dual-domain fusion convolutional neural network to fuse the features of pixel and histogram domains for contrast enhancement forensics. Specifically, we first present a pixel-domain convolutional neural network to automatically capture the patterns of contrast-enhanced images in the pixel domain. Then, we present a histogram-domain convolutional neural network to extract the features in the histogram domain. The feature representations of pixel and histogram domains are fused and fed into two fully connected layers for the classification of contrast-enhanced images. Experimental results show that the proposed method achieves better performance and is robust against pre-JPEG compression and antiforensics attacks, obtaining over 99% detection accuracy for JPEG-compressed images with different QFs and antiforensics attack. In addition, a strategy for performance improvements of CNN-based forensics is explored, which could provide guidance for the design of CNN-based forensics tools.     

Benchmarking Analysis of the Accuracy of Classification Methods Related to Entropy

In the machine learning literature we can find numerous methods to solve classification problems. We propose two new performance measures to analyze such methods. These measures are defined by using the concept of proportional reduction of classification error with respect to three benchmark classifiers, the random and two intuitive classifiers which are based on how a non-expert person could realize classification simply by applying a frequentist approach. We show that these three simple methods are closely related to different aspects of the entropy of the dataset. Therefore, these measures account somewhat for entropy in the dataset when evaluating the performance of classifiers. This allows us to measure the improvement in the classification results compared to simple methods, and at the same time how entropy affects classification capacity. To illustrate how these new performance measures can be used to analyze classifiers taking into account the entropy of the dataset, we carry out an intensive experiment in which we use the well-known J48 algorithm, and a UCI repository dataset on which we have previously selected a subset of the most relevant attributes. Then we carry out an extensive experiment in which we consider four heuristic classifiers, and 11 datasets.     

Laser self-mixing interference fibre sensor

Fibre sensors exhibit a number of advantages over other sensors such as high sensitivity, electric insulation, corrosion resistance, interference rejection and so on. And laser self-mixing interference can accurately detect the phase difference of feedback light. In this paper, a novel laser self-mixing interference fibre sensor that combines the advantages of fibre sensors with those of laser self-mixing interference is presented. Experimental configurations are set up to study the relationship between laser power output and phase of laser feedback light when the fibre trembles or when the fibre is stretched or pressed. The theoretical analysis of pressure sensors based on laser self-mixing interference is indicated to accord with the experimental results.     

Human eye localization using the modified Hough transform

The precise localization of parts of a human face such as mouth, nose or eyes is important for their image understanding and recognition. The developed successful computer method of eyes and eyelids localization using the modified Hough transform is presented in this paper. The efficiency of this method was tested on two publicly available face images databases and one private face images database with the location correctness better than 96% for a single eye or eyelid and 92% for eye and eyelid couples.     

飞秒激光制作微型光纤干涉传感器的研究进展

飞秒激光脉冲持续时间短、峰值功率高,在对物质进行处理过程中热作用区域小、加工精度高,已成为一种制作高性能、新结构光子器件的重要方法。简单介绍了飞秒脉冲激光微加工的方法和特性,归纳了飞秒激光直写技术在光纤干涉型传感器制作方面的研究进展,包括微型光纤Fabry-Perot干涉传感器和微型光纤Mach-Zehnder干涉传感器,重点介绍了这些光纤传感器的结构、特性和潜在应用。对用该技术制作的微型光纤干涉传感器进行了总结和展望。     

Thermal wave interference

Thermal waves are used in nondestructive testing for defect analysis by lock-in-thermography. A thermal wave field contains useful information about material and component properties, such as layer thickness or reflection coefficients of internal boundaries. Only the exact knowledge of the thermal wave field enables quantitative evaluation of these parameters. For example, a calculated wave field can be used for deconvolution of lock-in-thermography images as it provides a point spread function and therefore gives a better lateral resolution. For a better knowledge of thermal wave fields, basic experiments concerning interference effects are presented in this paper.     

Scaling Analysis of an Image Encryption Scheme Based on Chaotic Dynamical Systems

Despite that many image encryption systems based on chaotic or hyperchaotic systems have been proposed to protect different kinds of information, it has been crucial to achieve as much security as possible in such systems. In this sense, we numerically implement a known image encryption system with some variants, making special emphasis when two operations are considered in the scrambling stage. The variants of such an encryption system are based on some hyperchaotic systems, which generated some substitution boxes and the keys of the system. With the aim to have a more complete evaluation, some internal stages of the image encryption scheme have been evaluated by using common statistical tests, and also the scaling behavior of the encrypted images has been calculated by means of a two-dimensional detrended fluctuation analysis (2D-DFA). Our results show that the image encryption systems that include two operations or transformations in the scrambling stage present a better performance than those encryption systems that consider just one operation. In fact, the 2D-DFA approach was more sensitive than some common statistical tests to determine more clearly the impact of multiple operations in the scrambling process, confirming that this scaling method can be used as a perceptual security metric, and it may contribute to having better image encryption systems.     

A Quantum Private Query Protocol for Enhancing both User and Database Privacy

In order to protect the privacy of query user and database, some QKD-based quantum private query(QPQ)protocols were proposed. Unfortunately some of them cannot resist internal attack from database perfectly; some others can ensure better user privacy but require a reduction of database privacy. In this paper, a novel two-way QPQ protocol is proposed to ensure the privacy of both sides of communication. In our protocol, user makes initial quantum states and derives the key bit by comparing initial quantum state and outcome state returned from database by ctrl or shift mode instead of announcing two non-orthogonal qubits as others which may leak part secret information. In this way,not only the privacy of database be ensured but also user privacy is strengthened. Furthermore, our protocol can also realize the security of loss-tolerance, cheat-sensitive, and resisting JM attack etc.     

Multifocus image fusion and denoising scheme based on homogeneity similarity

A novel image fusion algorithm based on homogeneity similarity is proposed in this paper, aiming at solving the fusion problem of clean and noisy multifocus images. Firstly, the initial fused image is acquired with one multiresolution image fusion method. The pixels of the source images, which are similar to the corresponding initial fused image pixels, are considered to be located in the sharply focused regions. By this method, the initial focused regions are determined. In order to improve the fusion performance, morphological opening and closing are employed for post-processing. Secondly, the homogeneity similarity is introduced and used to fuse the clean and noisy multifocus images. Finally, the fused image is obtained by weighting the neighborhood pixels of the point of source images which are located at the focused region. Experimental results demonstrate that, for the clean multifocus image fusion, the proposed method performs better than some popular image fusion methods in both subjective and objective qualities. Furthermore, it can simultaneously resolve the image restoration and fusion problem when the source multifocus images are corrupted by the Gaussian white noise, and can also provide better performance than the conventional methods.