Computational Privacy Cameras

Major advances in computer vision and the mobile revolution have set the stage for the widespread deployment of connected cameras. This will lead to increased concerns about privacy and security. Conventional privacy solutions process sensitive data at the software level, which makes them susceptible to attacks. We develop computational privacy cameras that limit access to sensitive data by performing privacy processing prior to, during, and/or immediately after image capture, via application-specific integrated circuits (ASICs) and/or specialized optics. This will expand the range of places and personal devices where connected cameras can be deployed.

PAMI 2017, CVPR 2015 ICCP 2016
Pre-capture Privacy for Small Vision Sensors Sensor-level Privacy for Thermal Cameras
Francesco Pittaluga, Sanjeev J. Koppal Francesco Pittaluga, Aleksandar Zivkovic, Sanjeev J. Koppal
We enforce privacy at the optics-level, via privacy preserving optics that filter or block sensitive information directly from the incident light-field before sensor measurements are made, adding a new layer of privacy. In addition to balancing the privacy and utility of the captured data, we address trade-offs unique to miniature vision sensors, such as achieving high-quality field-of-view and resolution within the constraints of mass and volume. Our privacy optics enable applications such as depth sensing and full-body motion tracking. We enforce privacy at the sensor-level, as incident photons are converted into an electrical signal and then digitized into image measurements. By manipulating the sensor processes of gain, digitization, exposure time, and bias voltage, we are able to provide privacy during the actual image formation process and the original face data is never directly captured or stored. We show privacy-preserving thermal imaging applications such as temperature segmentation, night vision, gesture recognition and HDR imaging.
Paper (PAMI 2017)
Paper (CVPR 2015)
Paper (ICCP 2016)