SBIR Phase II: Novel photographic steganography to reduce digital piracy on live streaming platforms

The broader impact of this Small Business Innovation Research (SBIR) Phase II project will be to reduce video and image copyright-related liability for social media companies and retain profits for digital media licensing companies and other content owners/distributors who are holding or enforcing content rights. Beyond content protection, the technology being developed can also address the emerging need for digital content authentication that can assure users the digital images or videos they are viewing are legitimate and not altered through the use of “deep-fake” or other similar technologies. Global piracy of digital video has been estimated to cost media providers, content creators, and legal distributors $67-$79 billion in lost revenue. While companies pay billions of dollars per year for sports broadcast rights, illegal sports streaming sites can generate enormous ad revenues at virtually no cost. With the advent of the internet and the abundance of high-quality recording equipment on small, portable devices, it has become easier for video pirates to capture and share copyrighted material. Using photographic steganography to mark copyrighted content with undetectable messages will provide stakeholders with built-in anti-piracy protections. This technology, in turn, will help ensure legal content distribution, content protection for producers, and content assurance for consumers. This Small Business Innovation Research (SBIR) Phase II project seeks to develop a competitive solution to copyright protection through imperceptible watermarking of digital images and digital videos using photographic steganography. This project is distinct from prior work in that it: (1) models the human vision system so that pixel modulations are machine-readable but imperceptible to humans; (2) robustly handles video compression algorithms like H.264, HEVC, MPEG, and VP9; (3) employs single-frame, synchronization-free methodology compatible with both still images and videos; (4) enables free-space light communication (light field messaging) using smartphone cameras; (5) works with ordinary camera and display hardware with no special spectrum or speed requirements; and (6) employs state-of-the-art deep learning algorithms trained with hundreds of thousands of images. The proposed product will be able to reliably detect screen-to-camera piracy while maintaining broadcast video quality standards. During this project, the technology will be strengthened in four key technical areas of development: (i) robustness to critical image/video mutations; (ii) real-time encoding; (iii) computationally efficient decoding; and (iv) industry-standard certification for visual quality. This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.