New Compression Scheme for 3D Point Clouds Achieves Remarkable Results at Aggressive Bit RatesPublished on Mon Jul 24 2023 by Dustin Van Tate Testa PointCloud_CharlotteSt | Valerio on Flickr
Researchers have developed a new compression scheme for 3D point clouds, which are commonly used in applications like 3D scanning, virtual reality, and object detection. Point clouds require a large amount of data, making compression methods essential. The proposed method combines information on the geometrical features of the point cloud and the user's position to achieve remarkable results for aggressive compression schemes demanding very small bit rates.
The key innovation in this compression scheme is the use of saliency maps. After separating visible and non-visible points, four saliency maps are calculated, taking into account the geometry and distance from the user, visibility information, and the user's focus point. These maps indicate the overall significance of each point, allowing for the quantization of different regions with a different number of bits during the encoding process. The decoder then reconstructs the point cloud using delta coordinates and solving a sparse linear system.
Evaluation studies and comparisons with existing compression methods demonstrate that this new scheme achieves significantly better results for small bit rates. It achieves high compression ratios while preserving geometrically meaningful perceptible areas. The quality reduction in less significant parts of the scene is not noticeable to the user, even for aggressive compression rates. The method is ideal for applications that require extremely high compression rates and good perceptual accuracy.
Future research could focus on reducing execution time at the decoder, allowing for real-time performance on commodity hardware. Overall, this new compression scheme for 3D point clouds has the potential to greatly improve the efficiency and visual accuracy of applications that rely on these representations.