To compress the large amount of point cloud data is emerging as the dire need for the visualization of scientific simulation and for rendering biomedical information. This work proposes a novel technique for the compression of point cloud volumetric and surface data. It is based upon Multistage Vector Quantization (MSVQ) which is improvised for its application over 3D data. The clustering or initial codebook is generated with the help of hybridizing K-means clustering and Grow and Learn algorithm. The number of codevectors is determined with rate distortion constraint. Conclusively rate distortion analysis is also conducted for critical analysis of the algorithm.