Progressive Out-of-Core Compression Based on Reconstruction with Marching Cubes

Out-of-Core models are the massive models that cannot be loaded into the memory as a whole. For improving the efficiency of storing, transmitting and rendering such models, it is very important to progressively compress the models. However, to our knowledge, all the existing out-of-core compression algorithms are single rate, which cannot perform progressive compression. With regard to this, this paper proposes a method that can progressively compress out-of-core models in high efficiency and transmit and render the models in multi-resolutions. At first, the method uniformly divides the bounding box of the out-of-core model into sub-boxes for the local model in every sub-box to be able to process in core, and manages the sub-boxes hierarchically in an octree. Afterwards, the local model in every sub-box is reconstructed with Marching Cubes and an adaptable sub-octree is constructed for the reconstructed local model. Finally, based on the sub-octrees, the nodes of the octree of the whole model can be traversed progressively from coarse to fine and compressed with an advanced compression method for handling in-core models. Experimental results show that the new method can compress the out-of-core models in similar compression ratios as the advanced method to compress the in-core models,it is also superior to the existing methods for compressing out-of-core models and is the first method for performing progressive compression on out-of-core models.