Highly efficient computer oriented octree data structure and neighbours search in 3D GIS

Three-dimensional (3D) visualization has given a new perspective in various fields such as urban planning, hydrology, infrastructure modelling and geology. This is due to its capability of handling real world object in more realistic manners, rather than the two-dimensional (2D) approach. However, implementation of 3D spatial analysis in the real world situations has proven to be difficult to comprehend due to the complexity of the algorithm, computational process and time consuming. The existing Geographical Information Systems (GIS) enable 2D and two-and-a-half-dimensional (2.5D) spatial datasets, but less capable of supporting 3D data structures. Recent development in Octree showed that more effort was given to improve the weakness of Octree in finding neighbouring nodes by using various address encoding scheme with specific rule like matrix, lookup table and arithmetic to eliminate the need of tree traversal. Therefore, the purpose of this paper is to propose a new method to speed up the neighbouring search by eliminating the needs of complex operation to extract spatial information from Octree by preserving 3D spatial information directly from the Octree data structure. This new method will be able to achieve O(1) complexity and utilizing Bit Manipulation Instruction 2 (BMI2) to speed up address encoding, extraction and voxel search 1000x compared to generic implementation