Research And Optimization Of Static Point Cloud Compression Algorithm

With the future popularization of VR devices,people are increasing persuing a more immersive visual expierence.As a main representation of 3D scenes,point cloud are regarded as one of the most critical content in the next generation of immersice media.Existing 3D laser scanning euipment can obtain high-precision point cloud models easily,while massive point cloud data provides users with detailed and realistic scene presentations,it also brings greater pressure on transmission bandwidth and storage space,which makes the research on the compression of the point cloud data very necessary.However,the currently commonly used static point cloud compression schemes do not fully consider the distribution characteristics of the point cloud model itself in the 3D space,and there is still room for improvement in its coding performance.in this paper,an adaptive trisoup geometric information compression algorithm based on rate-distortion is proposed for geometric information encoding of point cloud,and a region adaptive hierarchical transform optimization encoding algorithm based on point cloud distribution is proposed for attribute information encoding of point cloud,they both improve the compression performance of point cloud data.In view of the problem of the gap caused by the slice partition and the problem of geometric parameter selection in the trisoup geometric information compression algorithm based on the surface approximation method,the research is carried out.By analyzing the reasons for the influence of slice division on the trisoup reconstruction process,a slice division method suitable for the trisoup geometric information compression framework considering the division position is proposed.On this basis,this paper considers that the point cloud has different geometric characteristics in different regions,proposes an adaptive trisoup geometric information compression algorithm based on rate-distortion optimization,which combined with slice partition and rate-distortion model.The experimental results show that,compared with the G-PCC coding scheme,the adaptive trisoup geometric information compression algorithm based on rate-distortion optimization proposed in this paper has higher performance.In the point-to-point distortion metric(D1)of geometric information,the BD-rate was reduced by an average of 2.7%.For the region adaptive hierarchical transform coding algorithm,the problem of the order of the transfrom direction is studied.Through experiments,it is found that the point cloud attribute information coding performance is different under different transformation orders.On the basis,this paper proposes an optimized coding scheme considering the point cloud distribution based on the region adaptive hierarchical transform.In the optimized region adaptive hierarchical transform coding algorithm,the order of the transform direction that can obtain better coding performance is determined by the characteristic of the distribution of the point cloud which is estimated based on the normal vectors.The experimental results show that compared with the existing algorithm,the optimaized algorithm proposed in this paper improves the coding performance to a certain extent,the BD-rate on the YUV channel of the color attribute information is reduced by 0.5%,1.2%,2.1% respectively.