Future Directions in Privacy-Enhancing Video Surveillance
By Ankur Chattopadhyay and Donxay Rasavong
As today’s video surveillance technology continues to get smarter exponentially, the debate on privacy versus security keeps on getting strong momentum with new questions coming up [6, 7]. The terrorist attacks on September 11th, 2001 gave the United States Government the power to blur the line between privacy and security without any major pushback by the American public. In today’s digital era of social media platforms and smartphones, as surveillance mechanisms keep on evolving with technological advances in computer vision, person re-identification (ReID) [2, 6] has emerged as a new threat to privacy in video surveillance. In a way, it challenges the Fourth Amendment in the American Constitution.
2 Person Re-Identification
Re-identification represents a technique that can link individuals across disparate datasets . In the case of video surveillance, this involves taking a captured image of an individual and re-identifying that individual using images from a different camera view . Establishing this correspondence across camera views is useful in several scenarios with the most obvious being the ability to track individuals’ movements throughout the camera network in an effort to further
Figure 1: Re-identification can track an individual across camera views
improve and enhance surveillance capabilities. In ReID, a probe image of an unknown individual is compared against a gallery of candidate images (known individuals) as an attempt to identify a match. Visual recognition algorithms can be used to scan the probe image against a database of reference images collected over the public domain (for instance, social media), and can then be stored for later use. Such a database would consist of low-level identifying features for individuals captured on the surveillance camera in multiple cross-view shots, which contribute to a sizeable collection of visual descriptors with stamped date, time and location.
3 Threat Against Privacy
As explained, the process of re-identification can pose a potentially serious threat to privacy of law-abiding citizens captured on public surveillance. This threat exists inspite of existing privacy-enhancing video surveillance techniques, which handle the explicit identifying characteristics (like face) of subjects. However, today’s surveillance videos contain a lot of implicit information (low-level identifying features like attire, background, belongings, gait, hair-color, gender indicators, etc,), which opens up the prospects for possible misuse of the video contents in the form of location tracking and re-dentification, and may negate the benefits of the current privacy-enhancing surveillance technologies. With today’s advanced computing power and sophisticated resources, it would be plausible to exploit the ReID technique and use a surveillance driven image database for tracking the location as well as for re-identifying an individual. Smart, advanced computer vision applications like ReID provide the abilities to gather implicitly identifying visual descriptors and to analyze them for engineering the ReID process, which can be a game changer in video surveillance. With lots of public information, including visual attributes, in the open via various social media platforms and the re-identification process lurking around, it is quite possible to learn about a given individual, recognize (or identify) and locate the person, thereby jeopardizing the individual’s private space.
3.2 Government Surveillance
Many government agencies have implemented machine learning techniques to process biometric databases and to exploit computer vision based machine learning’s “ability to mine visual data to get valuable insights about what’s happening in the world” . The Federal Bureau of Investigation (FBI) uses ReID through their Next Generation Identification (NGI) System, which contains “more complete and accurate identity records” and represents an “enhanced biometric identification repository” . For instance, with all the cameras used at tollbooths and intersections, the NGI System can process captured images of the passing vehicle and obtain the driver’s identity through visual recognition scans. With hundreds of thousands of images processed daily, the FBI can track criminals across state lines while storing data of innocent civilians. In addition, all the stored information from extensive footages can be exploited by other government entities resulting in compromise of privacy for national security reasons.
3.3 USA Freedom Act
The USA Freedom Act and its predecessor gave the federal government a wide range of power to collect data from private companies. A subpoena given to a company like Facebook would suffice to gain access to someone’s personal images. The federal government can use computer vision driven machine learning techniques to analyze the images and add details to complete the individual’s identity. The subpoena does not give the user an opportunity to know in advance that the government is looking through personal images, and this infringes the right to privacy.
3.4 Act of Congress
In addition, the United States Congress introduced Bill S 1872, also known as the Transportation Security Administration (TSA) Modernization Act, which would enable the TSA agency to use facial recognition technology at airports. The bill allows TSA to “facilitate, if appropriate, the deployment of biometric technology at checkpoints, screening lanes, bag drop and boarding areas, and other areas where such deployment would enhance security and facilitate passenger movement.”. The TSA would be able to use visual recognition technology to potentially streamline and efficiently move passengers through the security checkpoints. It would save passengers from the waiting in long lines at the airport. In addition, the government would be able to store and process the individual face scans for future use. These databases would give the government the ability to keep track of all the passengers in conjunction with other surveillance sources outside the airports.
The use of intelligent and sophisticated visual recognition techniques in line with the re-identification process, which is driven by a detailed collection of surveillance database images, is a real threat to our privacy that is protected under the Fourth Amendment. As shown in the Hollywood movie “Enemy of the State”, the federal government has absolute power and control. For the sake of national security, they can create a huge biometric database as well as knowingly use it to further their agenda without
disclosing any information to the public. Additionally, such a comprehensive database aided by a powerful computing system would be a huge target for hackers and foreign governments to exploit and track anyone at any given moment, which significantly undermines privacy and security of the public. Thus, future research work in the area of privacy-enhancing video surveillance design should focus on addressing the described privacy-invading implications of the sophisticated re-identification algorithms.
- J L. Sweeney, “k-anonymity: A model for protecting privacy,” International Journal of Uncertainty, Fuzziness and Knowledge-Based Systems, vol. 10, no. 5, pp. 557-570, 2002.
- A. Bedagkar-Gala and S. K. Shah, “Editor’s Choice Article: A Survey of Approaches and Trends in Person Re-identification,” Image Vison Comput, vol. 32, no. 4, pp. 270-286, 2014.
- Varah, MotionDSP Dr. Sean. “Video: The Next Frontier for Big Data.” Wired, Conde Nast, 6 Aug. 2015, www.wired.com/insights/2014/09/video-big-data/.
- “Next Generation Identification (NGI).” FBI, FBI, 6 May 2016, www.fbi.gov/services/cjis/fingerprints-and-other-biometrics/ngi.
- Feeney, Matthew. “Senate Bill Paves Way for Airport Facial Scans.” Forbes, Forbes Magazine, 6 Oct. 2017, www.forbes.com/sites/matthewfeeney2/2017/10/06/senate-bill-paves-way-for-airport-facial-scans/#698287b94fe0.
- Chattopadhyay, Ankur., “Developing an Innovative Framework for Design and Analysis of Privacy Enhancing Video Surveillance.” PhD diss., University of Colorado Colorado Springs. Kraemer Family Library, 2016.
- Chattopadhyay and T. E. Boult, “Privacycam: a privacy preserving camera using uclinux on the blackfin dsp,” in Computer Vision and Pattern Recognition, 2007. CVPR’07. IEEE Conference on, 2007, pp. 1-8.
Dr. Ankur Chattopadhyay is an Assistant Professor in the Information and Computing Sciences (ICS) department at the University of Wisconsin, Green Bay. He has a Ph.D. in computer science from the University of Colorado. His research interests include information assurance and cybersecurity, privacy-enhancing computer vision and pattern recognition, image processing & analysis, and computer science plus security education. He has published and presented in international conferences like IEEE Security & Privacy, ACM SIGCSE, IEEE CVPR and IEEE FIE. Chattopadhyay has more than 16+ years of experience in both academics and industry. As an academician, his passion is innovating computer science plus cybersecurity education, conducting research and applying his research to benefit the society. His industry profile includes multiple roles such as IT analyst, software engineer and embedded systems engineer, having worked with Tata Consultancy Services for several years.
Donxay Rasavong is a graduate student at the University of Maryland pursuing a MS in Information Assurance. He received his BS in Computer Science with an emphasis in Information Assurance and Security from the University of Wisconsin, Green Bay. He is currently working at Schneider National doing process automation work. His research interests are in security and technology, and he loves reading.
Dr. Steve Jones joined the Center for Information and Communication Sciences faculty in August of 1998. He came to Ball State University (BSU) from completing his doctoral studies at Bowling Green State University where he served the Dean of Continuing Education developing a distance-learning program for the College of Technology’s undergraduate Technology Education program. Dr. Jones was instrumental in bringing the new program on board because of his technical background and extensive research in the distance-learning field.
Prior to coming to higher education, Dr. Jones spent over sixteen and a half years in the communication technology industry. He owned his own teleconnect, providing high-end commercial voice and data networks to a broad range of end users. Dr. Jones provided all the engineering and technical support for his organization that grew to over twenty employees and two and a half million dollars per year revenue. Selling his portion of the organization in December of 1994, Dr. Jones worked briefly for Panasonic Communications and Systems Company as a district sales manager providing application engineering and product support to distributors in a five-state area prior to starting doctoral studies.