Wiener Filter-based Color Attribute Quality Enhancement for Geometry-based Point Cloud Compression

Point cloud is an important format of 3D data and has been applied in many fields. How to compress point cloud efficiently is an essential research topic. We propose a quality enhancement algorithm for color attribute of 3D point cloud via Wiener filter. We first calculate the optimal coefficients of Wiener filter for each color component in the encode side. Then, we write the coefficients into bit stream selectively according to the rate distortion cost before and after performing the Wiener filter. Finally, the decoder will perform the Wiener filter based on the transmitted coefficients. Experimental results show that an average 0.113dB, 0.233dB, and 0.246dB PSNR gain, corresponding to -2.9%, -4.84%, -5.36% BD-Rates, can be achieved, for Luma, Cb, and Cr components, respectively, by implementing the proposed method into the geometry-based point cloud compression platform.