Motion Detection in Low ResolutionVideo Surveillance Data to Provide Personal Privacy

摘要

Privacy protection from surreptitious video recordings is an important societal challenge. We desire a computer vision system (e.g., a robot) that can recognize human activities and assist our daily life, yet ensure that it is not recording video that may invade our privacy. This paper presents a fundamental approach to address such contradicting objectives: human activity recognition while only using extreme low-resolution anonymized videos. Although extensive research on action recognition has been carried out using standard video cameras, little work has explored recognition performance at extremely low temporal or spatial camera resolutions. Reliable action recognition in such a "degraded" environment would promote the development of privacy-preserving smart rooms that would facilitate intelligent interaction with its occupants while mitigating privacy concerns. Privacy protection from unwanted video recordings. We want a camera system to recognize important events and assist human daily life by understanding its videos, but we also want to ensure that it is not intruding the user's or others' privacy. This leads to two contradicting objectives. More specifically, we want to (1) prevent the camera system from obtaining detailed visual data that may contain private information, desirably at the hardware-level. Simultaneously, we want to (2) make the system capture as much detailed information as possible from its video, so that it understands surrounding objects and ongoing events for surveillance, life logging, and intelligent services. This paper presents an algorithm for motion detection in low resolution images.