Efficient Join Processing for Complex Rasterized Objects

One of the most common query types in spatial database management systems is the spatial intersection join. Many state-of-the-art join algorithms use minimal bounding rectangles to determine join candidates in a first filter step. In the case of very complex spatial objects, as used in novel database applications including computer-aided design and geographical information systems, these one-value approximations are far too coarse leading to high refinement cost. These expensive refinement cost can considerably be reduced by applying adequate compression techniques. In this paper, we introduce an efficient spatial join suitable for joining sets of complex rasterized objects. Our join is based on a cost-based decompositioning algorithm which generates replicating compressed object approximations taking the actual data distribution and the used packer characteristics into account. The experimental evaluation on complex rasterized real-world test data shows that our new concept accelerates the spatial intersection join considerably.