Regensky, Andy

Andy Regensky, M. Sc.

Department of Electrical-Electronic-Communication Engineering
Chair of Multimedia Communications and Signal Processing (Prof. Dr. Kaup)

Room: Room 06.0192
Cauerstr. 7
91058 Erlangen

Office hours

n.V.

Research

Since February 2020 I am dealing with the efficient compression of fisheye- and 360°-video data. Special focus is put on effective inter-prediction, where the current image is predicted based on already decoded, temporally preceding or following images. Previous methods for inter-prediction are designed for videos with perspective projection, which follow the pinhole camera model. The special projections of fisheye and 360° images, which can no longer be described by the pinhole camera model, lead to a significantly reduced efficiency of these methods. By incorporating the projection function into the process of inter-prediction, its effectiveness can be increased significantly. Thereby, the projection-based approach is not limited to inter-prediction and can potentially be applied to many other techniques in video coding and signal processing.

More information on the broad topic of next generation video communications at the chair can be found here: Next generation video communications

Theses Offered

I offer theses in the field of projection-based fisheye and 360° video coding at any time. Special attention is placed on sufficient experience with Python or C++ and basic knowledge of image and video signal processing. Of course, the required experience scales corresponding to previous study experience: Less experience is expected for a bachelor thesis than for a research internship or a master thesis. If you are interested, just send me an email with a short introduction of your interests and, if applicable, your previous experience in software development and image and video signal processing (e.g. lectures, internships, work experience, private projects).

Finished Theses

  • Generation of 3D Models from 360° Images
  • Coding of Distortion-Corrected Fisheye Video Sequences using H.266/VVC
  • Development of a Projection-based Intra Prediction Method for Fisheye Images
  • Improving Resampling in Viewport-Adaptive Fisheye Motion Compensation
  • Sphärische Bewegungsmodelle für die Blockbasierte Bewegungskompensation von 360°-Video
  • Investigation of a Rendering Pipeline for the Automatic Generation of Training Data for Checkerboard Segmentation
  • An Overview of the State of the Art in 360-Degree Image and Video Stitching
  • Comparison of the Resampling Quality for Non-Perspective Image Formats
  • Referenzfreie Brennweitenschätzung für perspektivische Videosequenzen

Awards

  • : Luise Prell-Preis (Luise Prell Stiftung) – 2021

Publications

2022

2021

2020

2019