RDO-based joint bit allocation for MPEG G-PCC

Abstract

In this paper, we propose a rate-distortion optimization (RDO) model for finding the joint bit allocation of geometry data and color data based on geometry-based point cloud compression (G-PCC) of Moving Picture Experts Group (MPEG). Currently, point cloud compression (PCC) is widely used in various fields such as artificial intelligence, augmented reality, and virtual reality. With 3D data acquisition technology development, the resulting point clouds are often very dense, resulting in a significant amount of time consumed for subsequent storage, transmission, and reconstruction. Therefore, it becomes more critical to maintain the compression quality while reducing the time. RDO is a method that has been maturely applied to find a compromise between reconstruction quality and the size of encoded bits under video coding. Nevertheless, RDO technology for PCC has not been the focus of research. Therefore, it is proposed to implement efficient compression by introducing RDO technology based on G-PCC. In this work, the RDO approach obtains the RD model by analyzing the geometry and color information in the training data. After that, the two rate-distortion (RD) models and the decision of the parameter \lambda are integrated to obtain the final RDO model. The experimental results were obtained by comparing common test conditions (CTC) based on MPEG G-PCC TMC13 v12.0 software. The experimental results show that this proposed method can reduce the geometry Bjøntegaard delta bitrate by 20% and increase the color Bjøntegaard delta bitrate by 37%.