Point cloud compression

Point cloud are among the most versatile and widely-adopted 3D modelling formats as they can be generated using different algorithms or devices, e.g., LiDAR, ToF cameras, SfM algorithms, laser scanners, and more. The compression and transmission of static and dynamic point clouds is a challenging issue that requires new efficient coding strategies because of the large amount of data, the sparsity of 3D points, and their noisy nature. In our research, we designed several strategies involving classical coding techniques (based on volume partitioning, non linear transform, entropy coding), as well as learned compression strategies (using autoencoders, adversarial architectures and neural implicit representation (NIR). The latter can also be combined with denoising and rendering approaches that can be tailored to immersive application.

Key research topics include:

• static point cloud compression
• dynamic point cloud compression
• LiDAR data transmission
• point cloud rendering for immersive applications VR/AR.

Selected publications:

@inproceedings{Capraro2019,

title = {Rendering-aware point cloud coding for mixed reality devices},

author = {Fabio Capraro and Simone Milani},

year  = {2019},

date = {2019-01-01},

urldate = {2019-01-01},

booktitle = {2019 IEEE International Conference on Image Processing (ICIP)},

pages = {3706--3710},

organization = {IEEE},

keywords = {},

pubstate = {published},

tppubtype = {inproceedings}

}

@inproceedings{Milani2017f,

title = {Fast point cloud compression via reversible cellular automata block transform},

author = {Simone Milani},

year  = {2017},

date = {2017-01-01},

urldate = {2017-01-01},

booktitle = {2017 IEEE International Conference on Image Processing (ICIP)},

pages = {4013--4017},

organization = {IEEE},

keywords = {},

pubstate = {published},

tppubtype = {inproceedings}

}