Compression of 3D Point Clouds Using a Region-Adaptive Hierarchical Transform.

Area of Science:

• Computer Graphics
• Video Compression
• Signal Processing

Background:

• Free-viewpoint video increasingly uses solid scene representations like point clouds over depth maps.
• Point clouds offer computational savings over meshes for real-time 3D graphics.
• Efficient compression of point cloud color data is crucial for real-time applications.

Purpose of the Study:

• To develop a novel method for compressing colors within 3D point clouds.
• To achieve high compression efficiency and computational performance for real-time 3D video.

Main Methods:

• A hierarchical sub-band transform, similar to an adaptive Haar wavelet, is applied to color data.
• Arithmetic coding with Laplace distributions models sub-band coefficients.
• Octree scanning is used for encoding point cloud geometry.
• A custom method transmits Laplace parameters for decoder reconstruction.

Main Results:

• The proposed color compression method achieves performance comparable to current state-of-the-art techniques.
• The solution demonstrates significantly higher computational efficiency.
• The approach is effective for intra-frame compression of point clouds.

Conclusions:

• The developed method represents a state-of-the-art solution for point cloud color compression in real-time 3D video.
• This technique offers a practical and efficient approach for handling dense 3D visual data.