Research - Point Cloud Coding

We are working, in the context of point cloud coding, on algorithms that enable their efficient compression and transmission. To this end, we are collaborating with the work group 7 of the ISO/IEC Moving Pictures Experts Group 3D Graphics Coding (3DG). As part of this work, we provide contributions to further develop current and future point cloud coding standards, like Video-based Point Cloud Coding (V-PCC) und Geometry-based Point Cloud Coding (G-PCC). Here are our specific research topics.

Graph-based 3D motion estimation and compensation

Point cloud sequences, which can be seen as "point cloud videos", are an increasingly relevant media type, which is, uncompressed, very costly to store and transmit. Just as with conventional videos, there is a lot of redundancy that can be exploited in order to reduce this cost. Motion estimation and compensation are of central importance when it comes to exploiting this redundancy. Since point clouds occupy the 3-dimensional domain they live in only very sparsely, processing them is a non-trivial task. That's why we investigate graphs as an alternative representation for point clouds, as well as their corresponding processing techniques.

Machine learning for efficient point cloud coding

The very promising results provided by neural networks and machine learning will be play a major role in point cloud coding. That's why we investigate efficient methods, based on both machine learning and traditional coding tools, and try to combine the best results of the two worlds.

Variable Video codec integration in V-PCC

One of the defining characteristic of V-PCC is that the standard must operate independently from the chosen underlying video coder. That's why we investigate the extent, by which the choice of the video codec (HEVC being currently used as a reference) improves the compression gain. More specifically, we study the influence of specific algorithms and tools within a given video coding standard, in order to optimize them for the compression of V-PCC videos.


The Institute's point cloud coding Group consists of Thibaut Meyer, M.Sc. and Dominik Mehlem, M.Sc..

Are you interested? There are many possible topics for bachelor and master theses in the area of point cloud coding. / .
Moreover, we offer student jobs in the field of point cloud coding, which are described at Student jobs.