Boosting MBR Based kNN Search Over Multimedia Data by Approximate Pruning Metric

MBR (Minimum Bounding Rectangle) has been widely used to represent multimedia data objects in R*-Tree family indexing techniques. In this paper, in order to improve the performance of kNN searching over multimedia data, we propose an approach to reduce the computation cost of MINMAXDIST by using its approximate upper bound instead of its precise value, and then we use it to construct two stronger heuristics for kNN pruning, which are helpful to avoid visiting unnecessary data objects and MBRs. The experimental results show that the proposed approach can reduce the computation cost and boost the overall performance in R*-Tree based kNN searching tasks.