Exploiting Data Coherency in Multiple Dataset Visualization

The paper deals with visualization of multiple datasets by exploiting data coherency among the datasets under consideration to address the issues related to storage and computation. The proposed data coherency approach utilizes the similarities among datasets by using the already generated polygon data for one or more reference datasets to approximate the isosurfaces of similar regions in other datasets. For finding the similarity among the multiple datasets, we use the octree data structure and compare datasets block by block. Blocks of a nonreference dataset, whose difference from the corresponding blocks of the reference dataset is within the user defined tolerance level, we use the already extracted polygon data for the reference data blocks to approximate the isosurfaces. Thus, only those non-reference data blocks, which differ from the reference blocks, are processed. Thereby, we reduce the number of polygons to represent the isosurfaces. To overcome the problem of cracks in thus extracted isosurfaces, we use a simple approach of overlapping at the interface between the directly processed and approximated portions of the isosurfaces. We have explored the effects of various factors including the data coherency condition, tolerance level and block size on the performance. Our results show that the proposed data coherency technique considerably improves the overall performance.