Anonymous subject identification and privacy information management in video surveillance

The widespread deployment of surveillance cameras has raised serious privacy concerns, and many privacy-enhancing schemes have been recently proposed to automatically redact images of selected individuals in the surveillance video for protection. Of equal importance are the privacy and efficiency of techniques to first, identify those individuals for privacy protection and second, provide access to original surveillance video contents for security analysis. In this paper, we propose an anonymous subject identification and privacy data management system to be used in privacy-aware video surveillance. The anonymous subject identification system uses iris patterns to identify individuals for privacy protection. Anonymity of the iris-matching process is guaranteed through the use of a garbled-circuit (GC)-based iris matching protocol. A novel GC complexity reduction scheme is proposed by simplifying the iris masking process in the protocol. A user-centric privacy information management system is also proposed that allows subjects to anonymously access their privacy information via their iris patterns. The system is composed of two encrypted-domain protocols: The privacy information encryption protocol encrypts the original video records using the iris pattern acquired during the subject identification phase; the privacy information retrieval protocol allows the video records to be anonymously retrieved through a GC-based iris pattern matching process. Experimental results on a public iris biometric database demonstrate the validity of our framework.     

基于多任务CNN的监控视频中异常行人快速检测

在近年来社会公共安全受到广泛关注的情况下,如何利用监控视频对异常行人进行监督,预防危险事件的发生成为了一个热门课题.异常行人是指与普通行人在外观上有明显异常性区别的人,例如用头盔大面积遮挡面部或低头躲避摄像头,考虑到异常行人的特征主要集中在头面部,本文提出一种基于多任务卷积神经网络和单类支持向量机的针对头面部特征的异常行人快速检测方法.首先进行头面部区域的检测,然后使用多任务卷积神经网络提取头面部区域的特征,之后使用单类支持向量机判断是正常行人还是异常行人.此外,本文还针对卷积神经网络设计了一种卷积核拆分方法,加快了特征提取的速度,最终实验表明,本文提出的算法能够快速有效的检测出监控视频中的异常行人.     

时空图卷积网络在人体异常行为识别中的应用

在监控场景下,由于监控资源短缺,行人异常行为容易发生漏检。针对该问题,提出了一种视频监控场景下的人体异常行为识别的方法,辅助监控人员及时发现异常。使用OpenPose对图像中行人进行人体骨架提取。针对图卷积网络对关节点特征聚合方式单一的问题,融合了基于图注意力网络（graph attention network,GAT）的图注意力机制。在改进后的图卷积网络的基础上,利用时空图卷积神经网络（spatial temporal graph convolutional networks,ST-GCN）,对行人关节点信息进行异常行为识别。实验结果表明,提出的识别算法对定义的行为识别准确率达85.48%,能够准确地识别监控视频中行人的异常行为。     

Understanding trust in privacy-aware video surveillance systems

Recent advances in pervasive video surveillance systems pave the way for a comprehensive surveillance of every aspect of our lives, hence, leading us to a state of dataveillance. Computerized and interconnected systems of cameras could be used to profile, track and monitor individuals for the sake of security. Notwithstanding, these systems clearly interfere with the fundamental right of the individuals to privacy. Most literature on privacy in video surveillance systems concentrates on the goal of detecting faces and other regions of interest and in proposing different methods to protect them. However, the trustworthiness of those systems and, by extension, of the privacy they provide are mostly neglected. In this article, we define the concept of trustworthy privacy-aware video surveillance system. Moreover, we assess the techniques proposed in the literature according to their suitability for such a video surveillance system. Finally, we describe the properties that a deployment of a trustworthy video surveillance system must fulfill.     

Preserving privacy by de-identifying face images

In the context of sharing video surveillance data, a significant threat to privacy is face recognition software, which can automatically identify known people, such as from a database of drivers' license photos, and thereby track people regardless of suspicion. This paper introduces an algorithm to protect the privacy of individuals in video surveillance data by deidentifying faces such that many facial characteristics remain but the face cannot be reliably recognized. A trivial solution to deidentifying faces involves blacking out each face. This thwarts any possible face recognition, but because all facial details are obscured, the result is of limited use. Many ad hoc attempts, such as covering eyes, fail to thwart face recognition because of the robustness of face recognition methods. This work presents a new privacy-enabling algorithm, named k-Same, that guarantees face recognition software cannot reliably recognize deidentified faces, even though many facial details are preserved. The algorithm determines similarity between faces based on a distance metric and creates new faces by averaging image components, which may be the original image pixels (k-Same-Pixel) or eigenvectors (k-Same-Eigen). Results are presented on a standard collection of real face images with varying k.     

基于时空共现模式的视觉行人再识别EI北大核心CSCD

基于视频图像的视觉行人再识别是指利用计算机视觉技术关联非重叠域摄像头网络下的相同行人,在视频安防和商业客流分析中具有重要应用.目前视觉行人再识别技术已经取得了相当不错的进展,但依旧面临很多挑战,比如摄像机的拍摄视角不同、遮挡现象和光照变化等所导致的行人表观变化和匹配不准确问题.为了克服单纯视觉匹配困难问题,本文提出一种结合行人表观特征跟行人时空共现模式的行人再识别方法.所提方法利用目标行人的邻域行人分布信息来辅助行人相似度计算,有效地利用时空上下文信息来加强视觉行人再识别.在行人再识别两个权威公开数据集Market-1501和DukeMTMC-ReID上的实验验证了所提方法的有效性.     

User-centric privacy awareness in video surveillance

Especially in urban environments, video cameras have become omnipresent. Supporters of video surveillance argue that it is an excellent tool for many applications including crime prevention and law enforcement. While this is certainly true, it must be questioned if sufficient efforts are made to protect the privacy of monitored people. Privacy concerns are often set aside when compared to public safety and security. One reaction to this situation is emerging: community-based efforts where citizens register and map surveillance cameras in their environment. Our study is inspired by this idea and proposes a user-specific and location-aware privacy awareness system. Using conventional smartphones, users not only can contribute to the camera maps, but also use community-collected data to be alerted of potential privacy violations. In our model, we define different levels of privacy awareness. For the highest level, we present a mechanism that allows users to directly interact with specially designed, trustworthy cameras. These cameras provide direct feedback about the tasks that are executed by the camera and how privacy-sensitive data is handled. A hardware security chip that is integrated into the camera is used to ensure authenticity, integrity and freshness of the provided camera status information.     

W3-privacy: understanding what, when, and where inference channels in multi-camera surveillance video

Huge amounts of video are being recorded every day by surveillance systems. Since video is capable of recording and preserving an enormous amount of information which can be used in many applications, it is worth examining the degree of privacy loss that might occur due to public access to the recorded video. A fundamental requirement of privacy solutions is an understanding and analysis of the inference channels than can lead to a breach of privacy. Though inference channels and privacy risks are well studied in traditional data sharing applications (e.g., hospitals sharing patient records for data analysis), privacy assessments of video data have been limited to the direct identifiers such as people’s faces in the video. Other important inference channels such as location (Where), time (When), and activities (What) are generally overlooked. In this paper we propose a privacy loss model that highlights and incorporates identity leakage through multiple inference channels that exist in a video due to what, when, and where information. We model the identity leakage and the sensitive information separately and combine them to calculate the privacy loss. The proposed identity leakage model is able to consolidate the identity leakage through multiple events and multiple cameras. The experimental results are provided to demonstrate the proposed privacy analysis framework.     

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。     

基于宽容训练和隐私保护的快速监控视频检索模型

监控视频关键帧检索和属性查找在交通、安防、教育等领域具有众多应用场景,应用深度学习模型处理海量视频数据在一定程度上缓解了人力消耗,但是存在隐私泄露、计算资源消耗大、时间长等特点.基于上述场景,提出了一个面向大规模监控视频的安全、快速的视频检索模型.具体地,根据云端算力大、监控摄像头算力规模小的特点,在云端部署重量级模型,并使用所提出的宽容训练策略对其进行定制化知识蒸馏,将蒸馏后的轻量级模型部署在监控摄像头内,同时使用局部加密算法对图像敏感部分进行加密,结合云端TEE技术和用户授权机制,在极低资源消耗的情况下实现隐私保护.通过合理控制蒸馏策略的“容忍度”,能够较好地平衡摄像头视频输入阶段和云端检索阶段的耗时,在保证极高准确率的前提下,保证极低的检索时延.相比于传统检索方法,该模型具有安全高效、可伸缩、低延时的特点.实验结果显示,在多个公开数据集上,该模型相比于传统检索方法提供9×-133×的加速.     

Tracking pedestrians using local spatio-temporal motion patterns in extremely crowded scenes

Tracking pedestrians is a vital component of many computer vision applications, including surveillance, scene understanding, and behavior analysis. Videos of crowded scenes present significant challenges to tracking due to the large number of pedestrians and the frequent partial occlusions that they produce. The movement of each pedestrian, however, contributes to the overall crowd motion (i.e., the collective motions of the scene's constituents over the entire video) that exhibits an underlying spatially and temporally varying structured pattern. In this paper, we present a novel Bayesian framework for tracking pedestrians in videos of crowded scenes using a space-time model of the crowd motion. We represent the crowd motion with a collection of hidden Markov models trained on local spatio-temporal motion patterns, i.e., the motion patterns exhibited by pedestrians as they move through local space-time regions of the video. Using this unique representation, we predict the next local spatio-temporal motion pattern a tracked pedestrian will exhibit based on the observed frames of the video. We then use this prediction as a prior for tracking the movement of an individual in videos of extremely crowded scenes. We show that our approach of leveraging the crowd motion enables tracking in videos of complex scenes that present unique difficulty to other approaches.     

Design and implementation of a secure and trustworthy platform for privacy-aware video surveillance

Worldwide, thousands of video surveillance cameras record our daily activities. People are aware that video surveillance is deployed for the sake of security. However, the privacy of individuals would be endangered if the proper measures were not considered. Privacy-aware video surveillance has historically been addressed by proposals based on detecting individuals and other sensitive parts of the video and hiding them using a variety of techniques. In this paper, we present a comprehensive solution tackling video processing, video protection and management of the Information System. We claim that a video surveillance system can protect our safety and, at the same time, guarantee our privacy. We describe the design and implementation of a privacy-aware video surveillance platform that, in order to be trustworthy, accomplishes with the properties of high detection accuracy, real-time performance and protected video utility. We have tested the proposed platform, and we demonstrate the feasibility of our approach for privacy protection.     

基于人体关节点的低头异常行人检测

近年来,随着智能手机的快速发展,低头族行人在过马路时依然保持浏览手机的姿态,由此造成的交通事故时有发生。如何有效检测低头族成为了当下亟待解决的问题。现有的检测方法需要大量的真实低头异常的数据集,且最终结果存在识别精度不高、速度不尽人意的问题。基于此,提出了一种快速有效的低头异常行人检测方法,与现有方法的区别在于该方法是基于关节点而不是图像。首先设计了一种构造数据集的方法,在识别人体关节点的基础上,调整左右腕关节坐标来模拟行人手持电子设备的姿态,解决了数据集缺少且需要大量标注的问题;其次,提出复杂环境中高效检测行人异常行为的算法,对上述关节点坐标进行分类识别,充分利用手臂与头部信息来实现行人异常行为检测。实验证明,所提算法能够实现实时检测,且检测精度达到了94.08%,从而可以为视频监控、驾驶员、辅助驾驶以及自动驾驶系统提供必要的参考信息。     

基于运动目标分类的双向人流量统计算法

提出了一种在智能视频监控中基于运动目标分类的双向人流量统计算法.本文首先对运动目标进行检测和跟踪,根据检测出的运动目标团块经过预设计数线时的特征信息,把目标划分为非行人、单行人和多行人.对于多行人的情况,利用HOG和SVM对目标团块中的头肩进行检测,判断出多目标团块包含的行人数目.在人流量的统计中,借助于运动目标方向信息和目标团块所包含的行人数目信息,对经过场景预设计数线的行人进行进出双向的统计.本文算在建筑物通道口环境下的人流量统计中取得了较好的效果.     

基于情景记忆的运动小目标行人检测神经网络

从视觉场景中可靠地检测小目标行人对象是构建未来人工智能视觉系统的重要基础。由于运动小目标的视感尺寸小且纹理特征模糊,导致现有的传统行人目标检测方法难以应对。针对该问题,基于蝗虫视觉系统的神经结构特性,借助人类大脑内侧颞叶（MTL）情景记忆认知机理,提出一种适用于运动小目标行人检测的人工视觉神经网络（STPDNN）模型。所提出的神经网络包括两部分：突触前和突触后子网络。其中,突触前网络模拟蝗虫视觉系统加工处理视觉信号的神经机理,获得表征目标对象低阶特征的视觉运动线索;突触后网络从低阶视觉信号中提取出行人目标的情景记忆高阶信息,以实现对运动目标的偏好性响应。系统性的实验结果表明,提出的STPDNN可有效检测视觉场景中的运动小目标行人对象。该研究工作涉及生物视神经机理启发的行人目标动态视觉信息加工处理,可为智能视频监控中的行人检测识别与运动行为分析提供新思想、新方法。     

Surveillance camera scheduling: a virtual vision approach

We present a surveillance system, comprising wide field-of-view (FOV) passive cameras and pan/tilt/zoom (PTZ) active cameras, which automatically captures high-resolution videos of pedestrians as they move through a designated area. A wide-FOV static camera can track multiple pedestrians, while any PTZ active camera can capture high-quality videos of one pedestrian at a time. We formulate the multi-camera control strategy as an online scheduling problem and propose a solution that combines the information gathered by the wide-FOV cameras with weighted round-robin scheduling to guide the available PTZ cameras, such that each pedestrian is observed by at least one PTZ camera while in the designated area. A centerpiece of our work is the development and testing of experimental surveillance systems within a visually and behaviorally realistic virtual environment simulator. The simulator is valuable as our research would be more or less infeasible in the real world given the impediments to deploying and experimenting with appropriately complex camera sensor networks in large public spaces. In particular, we demonstrate our surveillance system in a virtual train station environment populated by autonomous, lifelike virtual pedestrians, wherein easily reconfigurable virtual cameras generate synthetic video feeds. The video streams emulate those generated by real surveillance cameras monitoring richly populated public spaces.A preliminary version of this paper appeared as [1].     

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.     

基于深度神经网络的行人头部检测

行人检测已成为安防、智能视频监控、景区人流量统计所依赖的核心技术,最新目标检测方法包括快速的区域卷积神经网络Fast RCNN、单发多重检测器 SSD、部分形变模型DPM等,皆为对行人整体的检测。在大场景下,行人姿态各异,物体间遮挡频繁,只有通过对行人身体部分位置建模,抓住人的局部特征,才能实现准确的定位。利用Faster RCNN深度网络原型,针对行人头部建立检测模型,同时提取行人不同方向的头部特征,并加入空间金字塔池化层,保证检测速率,有效解决大场景下行人的部分遮挡问题,同时清晰地显示人群大致流动方向,相比普通的人头估计,更有利于人流量统计。     

Protecting the privacy of humans in video sequences using a computer vision-based de-identification pipeline

We propose a computer vision-based de-identification pipeline that enables automated protection of privacy of humans in video sequences through obfuscating their appearance, while preserving the naturalness and utility of the de-identified data. Our pipeline specifically addresses de-identifying soft and non-biometric features, such as clothing, hair, skin color etc., which often remain recognizable when simpler techniques such as blurring are applied. Assuming a surveillance scenario, we combine background subtraction based on Gaussian mixtures with an improved version of the GrabCut algorithm to find and segment pedestrians. De-identification is performed by altering the appearance of the segmented pedestrians through the neural art algorithm that uses the responses of a deep neural network to render the pedestrian images in a different style. Experimental evaluation is performed both by automated classification and through a user study. Results suggest that the proposed pipeline successfully de-identifies a range of hard and soft biometric and non-biometric identifiers, including face, clothing and hair.     

基于部位检测和子结构组合的行人检测方法

提出了一种基于部位检测和子结构组合的、可用于辅助驾驶或视频监控系统中行人检测的方法.首先使用头部分类器在整幅图像中检测,得到感兴趣区域;然后在每个感兴趣区域内使用头部、躯干、腿部以及左臂和右臂5个人体部位检测器分别检测并使用基于子结构的检测组合方法对部位检测结果进行组合,以得到最终结果.在不同数据库上的实验结果表明,本方法可以有效地用于移动或静止摄像机所拍摄的视频图像中的多姿态及部分遮挡的行人检测.