Privacy enabled video surveillance using a two state Markov tracking algorithm

Nowadays video surveillance systems are widely deployed in many public places. However, the widespread use of video surveillance violates the privacy rights of the people. Many authors have addressed the privacy issues from various points of view. In this paper we propose a novel, on-demand selectively revocable, privacy preserving mechanism. The surveillance video can be tuned to view with complete privacy or by revoking the privacy of any subset of pedestrians while ensuring complete privacy to the remaining pedestrians. We achieve this by tracking the pedestrians using a novel Markov chain algorithm with two hidden states, detecting the head contour of the tracked pedestrians and obscuring their faces using an encryption mechanism. The detected pedestrian face/head is obscured by encrypting with a unique key derived from a master key for the privacy preservation purpose. The performance evaluations on many challenging surveillance scenarios show that the proposed mechanism can effectively and robustly track as well as identify multiple pedestrians and obscure/unobscure their faces/head in real time.     

基于面部特征图对称的人脸正面化生成对抗网络算法

目前人脸正面化研究主要解决人脸偏转问题,而对监控视频等现实场景中同时受偏转和俯仰变化影响的侧脸的正面化生成关注较少,针对这个问题和多角度侧脸生成的正面人脸图存在身份信息保留不全的问题,提出了一种基于特征图对称模块和眼周特征保留损失的生成对抗网络（GAN）。首先,根据人脸对称性先验,提出特征图对称模块,先使用人脸关键点检测器检测出侧脸鼻尖点位置,再将编码器提取到的特征图依照鼻尖位置进行镜像对称,从而在特征层面上缓解面部信息缺失的问题。其次,借鉴眼周识别思想,在现有的生成图身份保留方法中加入了眼周特征保留损失以训练生成器生成逼真的且保留身份信息的人脸正面图像。实验结果表明,所提算法得到的生成图面部细节保留较好,且在CAS-PEAL-R1数据集的所有俯角下人脸的平均Rank-1识别率为99.03%,可见该算法能够有效解决多角度侧脸的正面化问题。     

视频监控下的人脸跟踪与识别系统

针对智能视频监控系统的要求,设计了一个基于视频监控的自动多人脸跟踪识别系统,该系统的功能是实时跟踪视频监控范围内的人脸并鉴别人脸的身份。针对复杂背景及类似人脸区域的影响,提出了一种Adaboost人脸检测算法和主动形状模型相结合的人脸检测算法,实现人脸的准确检测;针对视频监控范围内人脸偏转、交错以及由于人员不断出入而导致人脸数目发生变化的问题,提出了CamShift和Kalman滤波器相结合的多人脸跟踪算法,同时对跟踪到的人脸进行实时身份识别。实验证明,该系统在视频监控范围内对人脸检测和身份识别准确,跟踪实时性好,是一种建立实时视频监控系统的实用方法。     

一种自动鲁棒的三维人脸重建方法

三维人脸模型已经广泛应用到视频电话、视频会议、影视制作、电脑游戏、人脸识别等多个领域。目前三维人脸建模一般使用多幅图像，且要求表情中性。本文提出了基于正、侧面任意表情三维人脸重建方法。首先对二维图像中的人脸进行特征提取，然后基于三维人脸统计模型，通过缩放、平移、旋转等方法，及全局和局部匹配，获得特定的三维人脸。基于二维图像中的人脸纹理信息，通过纹理映射，获得完整的三维人脸。通过对大量实际二维人脸图像的三维人脸重建，证实了该方法的有效性和鲁棒性。     

基于改进CycleGAN的视频监控人脸超分辨率恢复算法

针对有监督超分辨率算法训练过程需要大量成对图像、处理真实低分辨率图像视觉恢复效果差等问题,提出了一种基于改进CycleGAN的半监督算法Cycle-SRNet.首先,利用退化模型获得与真实低分辨率人脸相似的图像,用于训练网络参数;其次,通过重建模型恢复出具有真实效果的高分辨率人脸图像;最后引入感知损失函数保持人脸结构相似性,以更好地恢复面部特征.实验结果表明,该算法不需要成对的图像进行网络训练,在视觉效果上能够将模糊的视频监控低分辨率人脸图像恢复成清晰可辨的人脸图像,在FID、PSNR和SSIM指标上超越了SRCNN、SRGAN、CinCGAN等方法.     

基于视频监控的人脸识别方法

目前人脸识别方法主要针对静态图像进行识别,而在监控视频中,不同视频帧人脸具有相关性且只有部分人脸能够有效反映人脸信息。根据监控视频中人脸图像变化特性,提出了一种基于监控视频的人脸识别方法。首先通过结合人脸检测与跟踪技术获得视频人脸序列,然后以视频人脸序列中部分人脸图像识别结果为导向选取全部人脸序列图像中的代表人脸图像进行识别,最后根据选取的全部人脸图像识别结果综合反映人脸信息。实验结果表明,该方法能够在确保识别率和误识率的前提下有效提升监控视频中人脸识别的实时性。     

Face detection based on multilayer feed-forward neural network and Haar features

Fast and accurate detection of a facial data is crucial for both face and facial expression recognition systems. These systems include internet protocol video surveillance systems, crime scene photographs systems, and criminals' databases. The aim for this study is both improvement of accuracy and speed. The salient facial features are extracted through Haar techniques. The sizes of the images are reduced by Bessel down-sampling algorithm. This method preserved the details and perceptual quality of the original image. Then, image normalization was done by anisotropic smoothing. Multilayer feed-forward neural network with a back-propagation algorithm was used as classifier. A detection accuracy of 98.5% with acceptable false positives was registered with test sets from FDDB, CMU-MIT, and Champions databases. The speed of execution was also promising. An evaluation of the proposed method with other popular detectors on the FDDB set shows great improvement.     

人脸样貌独立判别的协作表情识别算法

为了降低样貌、姿态、眼镜以及表情定义不统一等因素对人脸表情识别的影响,提出一种人脸样貌独立判别的协作表情识别算法。首先,采用自动的人脸检测算法定位、对齐视频每帧的人脸区域,并从人脸视频序列中选择峰值表情的人脸;然后,采用峰值人脸与某个表情类内的所有人脸产生表情类内差异人脸信息,并通过计算峰值表情人脸与表情类内差异人脸的差异信息获得协作的表情表示;最终,采用基于稀疏的分类器与表情表示决定每个人脸表情的标签。采用欧美与亚洲人脸的数据库进行仿真实验,结果表明本算法获得了较好的表情识别准确率,对不同样貌、佩戴眼镜的人脸样本也具有较好的识别效果。     

Dynamics of facial expression: recognition of facial actions and their temporal segments from face profile image sequences.

Automatic analysis of human facial expression is a challenging problem with many applications. Most of the existing automated systems for facial expression analysis attempt to recognize a few prototypic emotional expressions, such as anger and happiness. Instead of representing another approach to machine analysis of prototypic facial expressions of emotion, the method presented in this paper attempts to handle a large range of human facial behavior by recognizing facial muscle actions that produce expressions. Virtually all of the existing vision systems for facial muscle action detection deal only with frontal-view face images and cannot handle temporal dynamics of facial actions. In this paper, we present a system for automatic recognition of facial action units (AUs) and their temporal models from long, profile-view face image sequences. We exploit particle filtering to track 15 facial points in an input face-profile sequence, and we introduce facial-action-dynamics recognition from continuous video input using temporal rules. The algorithm performs both automatic segmentation of an input video into facial expressions pictured and recognition of temporal segments (i.e., onset, apex, offset) of 27 AUs occurring alone or in a combination in the input face-profile video. A recognition rate of 87% is achieved.     

TH FACE2000人脸识别系统的设计与实现

文章着重介绍了THFACE2000人脸自动识别系统的设计思想和实现技术。该系统以人像识别技术为核心,用客户机/服务器模式,建立和维护了一个含有人像特征的综合数据库,从而能够在输入一张未知人像照片时,迅速地询出最接近于待查照片的若干已登记人员的身份信息。该系统应用到了人像识别,机群,数据库等多种技术。该系统在安刑侦,海关,机场,车站等领域有着广泛的应用前景。     

A general framework for managing and processing live video data with privacy protection

Though a large body of existing work on video surveillance focuses on image and video processing techniques, few address the usability of such systems, and in particular privacy issues. This study fuses concepts from stream processing and content-based image retrieval to construct a privacy-preserving framework for rapid development and deployment of video surveillance applications. Privacy policies, instantiated to as privacy filters, may be applied both granularly and hierarchically. Privacy filters are granular as they are applicable to specific objects appearing in the video streams. They are hierarchal because they can be specified at specific objects in the framework (e.g., users, cameras) and are combined such that the disseminated video stream adheres to the most stringent aspect specified in the cascade of all privacy filters relevant to a video stream or query. To support this privacy framework, we extend our Live Video Database Model with an informatics-based approach to object recognition and tracking and add an intrinsic privacy model that provides a level of privacy protection not previously available for real-time streaming video data. The proposed framework also provides a formal approach to implement and enforce privacy policies that are verifiable, an important step towards privacy certification of video surveillance systems through a standardized privacy specification language.     

Automatic face detection in video sequences using local normalization and optimal adaptive correlation techniques

Automatic human face detection from video sequences is an important component of intelligent human computer interaction systems for video surveillance, face recognition, emotion recognition and face database management. This paper proposes an automatic and robust method to detect human faces from video sequences that combines feature extraction and face detection based on local normalization, Gabor wavelets transform and Adaboost algorithm. The key step and the main contribution of this work is the incorporation of a normalization technique based on local histograms with optimal adaptive correlation (OAC) technique to alleviate a common problem in conventional face detection methods: inconsistent performance due to sensitivity to variation illuminations such as local shadowing, noise and occlusion. The approach uses a cascade of classifiers to adopt a coarse-to-fine strategy for achieving higher detection rates with lower false positives. The experimental results demonstrate a significant performance improvement gains and achieved by local normalization over methods without normalizations in real video sequences with a wide range of facial variations in color, position, scale, and varying lighting conditions.     

人脸的层次化描述模型及识别研究

人脸自动识别是一个困难但有重要意义的工作。文中提出了一种基于人脸层次化描述的识别方法。该方法首先对人脸进行快速准确的特征定位及标准化,然后采用主元分析神经网络分别对定位的人脸及其特征区域进行最佳特征提取,从而得到人脸在低分辨率和较高分辨率上的两层特征描述用以识别,具有识别率高、特征数据量适中、可用于大量人像识别等特点。此方法在１３００幅人像上进行了测试,结果表明其在人脸转动、表情变化或入脸未经训练     

Discrete area filters in accurate detection of faces and facial features

This paper introduces a new method for detection of faces and facial features. Proposed algorithm denies the thesis that bottom-up solutions can't work at reasonable speed. It introduces fast detection – about 9 frames per second for a 384 × 256 image – while preserving accurate details of the detection. Main experiments focus on the detection of the eye centers — crucial in many computer vision systems such as face recognition, eye movement detection or iris recognition, however algorithm is tuned to detect 15 fiducial face points. Models were trained on nearly frontal faces. Bottom-up approach allows to detect objects under partial occlusion — particularly two out of four face parts (left eye, right eye, nose, mouth) must be localized. Precision of the trained model is verified on the Feret dataset. Robustness of the face detection is evaluated on the BioID, LFPW, Feret, GT, Valid and Helen databases in comparison to the state of the art detectors.     

基于视频监控的人脸检测跟踪识别系统研究

设计了一种基于视频监控的人脸检测跟踪识别系统,该系统的功能是检测并实时跟踪视频中的人脸图像,同时进行身份识别。针对Gentle AdaBoost算法构造的级联分类器检测效率偏低的问题,提出了一种递进复杂度的级联分类器。针对传统粒子滤波器最高权重粒子不准确的问题,提出了均值权重粒子滤波器。针对传统粒子滤波器样本衰退的问题,提出了一种同时结合人脸检测和人脸跟踪算法的跟踪校正策略。对于检测和跟踪到的人脸,利用基于Gabor变换和HMM的方法进行身份识别。实验结果表明,系统能够准确地检测并实时跟踪视频中的人脸,可以实现人脸的快速识别,是一种能够应用到视频监控系统中的有效方法。     

Robust Video Face Recognition Under Pose Variation

With the abundance of video data, the interest in more effective methods for recognizing faces from surveillance videos has grown. However, most algorithms proposed in this field have an assumption that each image set lies in a single linear subspace, or a mixture of linear subspaces. As a result, 3-dimensional shape information, which leads to the nonlinear transformation of face images, is ignored. This paper proposes a robust video face recognition across pose variation in video (RVPose) based on sparse representation. The key idea is performing alignment and recognition based on sparse representation simultaneously. Moreover, by considering that multi-pose faces of the same subject possess the same texture and 3-dimensional shape, RVPose aligns a sequence of faces with pose variations simultaneously, which is reduced to a 3-dimensional shape-constrained video alignment problem. Finally, aligned video sequence is recognized based on sparse represent. Experiments conducted on public video datasets demonstrate the effectiveness of the proposed algorithm.     

Faceless identification based on temporal strips

This paper first presents a novel approach for modelling facial features, Local Directional Texture (LDT), which exploits the unique directional information in image textures for the problem of face recognition. A variant of LDT with privacy-preserving temporal strips (TS) is then considered to achieve faceless recognition with a higher degree of privacy while maintaining high accuracy. The TS uses two strips of pixel blocks from the temporal planes, XT and YT, for face recognition. By removing the reliance on spatial context (i.e., XY plane) for this task, the proposed method withholds facial appearance information from public view, where only one-dimensional temporal information that varies across time are extracted for recognition. Thus, privacy is assured, yet without impeding the facial recognition task which is vital for many security applications such as street surveillance and perimeter access control. To validate the reliability of the proposed method, experiments were carried out using the Honda/UCSD, CK+, CAS(ME)² and CASME II databases. The proposed method achieved a recognition rate of 98.26% in the standard video-based face recognition database, Honda/UCSD. It also offers a 81.92% reduction in the dimension length required for storing the extracted features, in contrast to the conventional LBP-TOP.

     

基于人脸认知模式的相似脸搜索

目的 人类对人脸认知模式的探索由来已久,并且已经成功应用于美容整形等研究领域。然而,目前在计算机视觉和模式识别领域,计算人脸相似度的方法没有考虑人对人脸的认知模式,使得现有方法的计算结果从人的认知习惯角度来讲并非最佳。为克服以上缺陷,提出一种基于人脸认知模式的相似脸搜索算法。方法 依据人脸认知模式,选取特征点,并计算特征量,构造各面部器官（眼睛、鼻子、嘴巴、脸型）分类模型,即面部器官形状相似性度量模型,并采用圆形LBP算子,计算两幅人脸对应器官的纹理相似度,二者综合作为相似脸搜索的依据。结果 分别用本文方法和代表相似脸搜索最高水平的Face++的方法对80幅正面、中性表情、平视角度拍摄的人脸图像进行测试。本文方法的整体准确率高于Face++方法,其中,TOP1、TOP2最相似搜索结果准确率优势明显,均高出Face++方法12%以上。结论 实验结果表明,本文方法的搜索结果更加符合人脸认知模式,可应用于正面、中性表情、平视角度拍摄的人脸图像的相似脸搜索。此外,还可以将此类基于认知模式的图像搜索思路推广应用于商业领域,如基于图像的相似网购商品搜索等。     

Defeating face de-identification methods based on DCT-block scrambling

Face de-identification aims at preserving the privacy of people by concealing faces in images and videos. In this paper, we propose a defeating algorithm for face de-identification methods that are based on DCT-block scrambling. These methods protect faces by scrambling the AC and DC coefficients of the DCT blocks corresponding to a face region in the compressed domain. The proposed approach does not make use of the protection key utilized in the de-identification process. It consists of the following stages. First, random unprotected faces are generated based on a random alteration of the sign of AC coefficients with a fixed value of DC coefficients. Then, the best unprotected faces are selected by an eigenfaces model trained with facial images from a repository of potentially protected people. A single facial image is then generated by merging the selected images through median stacking. Finally, the eigenfaces model is utilized again to choose the face from the repository that is closest to the resulting image in order to improve the aspect of the unprotected face. Experimental results using a proprietary database and the public CALTECH, Utrecht and LFW face databases show the effectiveness of the proposed technique.     

人脸识别视频分析的设计与实现

智能视频监控随着大数据视频分析技术的应用得到迅速发展,通过实时监控,视频分析系统即时反应。识别出视频监控采集到的目标人脸。在基于ARM11的S3C6410嵌入式硬件平台,以Linux操作系统为软件平台,用LBP算法对人脸图像进行预处理,减少光照对识别的干扰;用PCA算法对经一致模式统计后的均衡直方图进行特征提取和降维,降低数据的计算量;对降维后的数据运用岭回归算法进行分类,设计并优化LBP＋PCA算法,结合RidgeRegression分类技术作为人脸识别视频分析的解决方案。