Point-Cloud compression using data independent method — A 3D discrete cosine transform approach
Lujia Wang, Luyu Wang, Yinting Luo, Ming Liu · 2017
Point-cloud is a widely used representation for objects and scenes. It generally consists of a large amount of 3D coordinates of points describing reflective surfaces. A subtle problem is that the number of points is usually so large that real-time transmission and efficient storage is not feasible. In this work, we propose to use 3D Discrete Cosine Transform (3D-DCT) to compress these two typical categories of data, namely point-cloud data extracted from objects and environments (i.e. 3D maps). Experimental results show that the proposed method leads to high compression ratio and flexible reconstruction behaviors comparing with other related methods.