We address the problem of efficient processing of count aggregate queries for spatial objects in OLAP systems. One of the main issues affecting the efficient spatial analysis is the, so called, distinct counting problem. The core of the problem is due to the fact that spatial objects such as lakes, rivers, etc... - and their representations - have extents. We investigate the trade-offs that arise when (semi) materialized views of the count aggregate are maintained in a hierarchical index and propose two data structures that are based on the Quadtree indexes: Fully Materialize Views (FMV) and Partially Materialized Views (PMV). Each aims at achieving a balance between the: (1) benefits in terms of response time for range queries; (2) overheads in terms of extra space and update costs. Our experiments on real datasets (Minnesota lakes) demonstrate that the proposed approaches are beneficial for the first aspect achieving up to five times speed-up, while incurring relatively minor overheads with respect to the second one, when compared to the naïve approach.