Preserving Privacy Using Visual Cryptography in Surveillance Systems

摘要

This study proposes a practical, cost-effective and privacy-enhancing visual cryptographic surveillance system solution to independently store multiple shares of the original camera images. A face-detection algorithm determines faces on the original real-time image, and then the determined faces are replaced with random data to protect people's identities. The rest of the image are kept, as it is to be able to follow the acts within the surveillance system. Whenever a suspected act occurs, independent entities in the system may act together to re-compose the original image and identify the suspected person. The proof of concept implementation, attack model and security analysis of the proposed model is discussed within the paper.