Ongoing Thesis
Master's Theses
Video Coding of Natural Video Sequences Using Estimated Auxiliary Data
video coding, codec, JVET
Description
As part of the AhG15 (gaming content compression) activity at JVET, additional data is available for various sequences: depth maps, motion vectors, camera parameters, and trajectories. These were directly extracted from a gaming engine, ensuring their superior quality. AhG15 aims to investigate how this data can be leveraged for the compression of gaming content.
In this master’s thesis, we want to explore whether similar data for regular sequences can also be estimated and utilized for coding. Auxiliary data can be obtained by leveraging existing approaches, such as SfM and SLAM. The final goal is to achieve higher coding gain (BD-Rate) or faster coding by simplifying the motion search process.
Supervisors: Dr. Johannes Sauer (Huawei Munich), Hongjie You
Prerequisites
Knowledge on video coding, H.266 (VVC) and H.265 (HEVC)
Coding skills of C++, python and shell for working with JVET AhG verification software.
Independent problem solving skills.
Contact
Please get in touch with Hongjie You directly