A model for the analysis of neighbor finding in pointer-based quadtrees

A natural byproduct of the tree-like nature of the quadtree is that many basic image processing operations can be implemented as tree traversals which differ in the nature of the computation that is performed at each node. Some of these computations involve the inspection of a node's adjacent neighbors (termed neighbor finding). A new model is developed for images represented by quadtrees, and it is used to analyze various neighbor-finding techniques. The model's predicted costs for neighbor finding correlate very closely with empirical results and it is superior to the model that was used previously.