Frédéric Lefebvre


Imaging Science Lab Researchers

During his PhD period, DrFrederic Lefebvre pointed out an innovative solution for content identification and indexing based on image hashing for multimedia content. His main contribution relates to perceptual hashing for images and video. Bit to bit comparison of conventional cryptographic hash values (MD5, SHA1) is doomed to fail providing reliable content identification. In contrast, the proposed one-way function for images (also referred to as perceptual hash, robust hash or semantic hash) extracts perceptually significant features from the content which are robust to content manipulation.

He is the author of more than 20 research articles in refereed international journals or conferences on content identification and copy protection topics and granted more than 10 patents.

His research focuses on content protection, content identification, content registration, selective encryption, color science and computer vision.

Once he joined Technicolor (Thomson) in 2005, first, he launched the video fingerprint activity inside Technicolor and took the leadership of the fingerprinting team. He has developed the first Video Fingerprint product for Technicolor. The Technicolor Video Fingerprint technique targets copy protection applications and aims at filtering illegal content over P2P network and User Generated Content (UGC). The main content identification challenges in copy protection are to provide an optimal tradeoff between speed, storage, and detection rate.

Then, he launched selective encryption for multimedia content, a new domain between security and compression. Selective encryption aims at reducing the amount of data to encrypt while achieving a sufficient and inexpensive security and keeping compliance to the compression standard. This approach is particularly desirable in constrained communication (real time networking with delay constraints, mobile communication with limited computational power...).

Finally, he decided to discover new areas in computer vision and color science. First, he contributed to a new product for the Business Units: color re mastering of legacy content. In this project, he developed new synchronization process (frame accurate)in order totransfer color properties (grading) from master SD contents to master HD content. Then, he proposeda new color transfer algorithms. Color transfer between images consists in modifying the colors in a source image such that it takes the color characteristics (color look) of an example image, while preserving the content of the source image. And now, he is working on Rotoscoping techniquesin order tohelp the colorists and artists to track and segment the object in real scene.

Video Fingerprinting, Image Video processing Watermarking, Selective encryption, Color Science and Computer Vision.

We use cookies on our website to support technical features that enhance your user experience.

We also use analytics & advertising services. To opt-out click for more information.