Conclusions

In this paper, a technique has been introduced for use in quadtree-based access methods to improve previously proposed methods and efficiently process window queries in thematic maps with multiple non-overlapping features. A variation of the Linear quadtree has been presented and algorithms to process window queries have been described. The results of the conducted experiments concerning the performance of the new method when applied to the window queries have been discussed. It has been shown that this new method has a stable behavior in all cases, performing either the best or close to the best.

Future work may focus on a more efficient clustering of quadrants with similar features. This could be achieved either with the use of signature trees (S-trees) [3] or two-dimensional R-trees, where the first dimension would be dedicated for feature information and the second one would be the locational code.