Compression of Large 3D Engineering Models using Automatic Discovery of Repeating Geometric Features

In this paper, we present a new geometry compression technique particularly suitable for 3D mesh models of engineering class – architectural models, machine plants, factories, etc. We observe that such models have a number of repeating features at various levels of granularity. In most of the widely available models in this class, the geometric description of such features are also repeated. A major distinctive aspect of our compression technique is that repeating geometric features in the geometric model are automatically discovered and then compactly encoded. The proposed method discovers repetition first at the connected component level and then at the subcomponent level across components and also at the aggregate component level. The results from a straight forward implementation tried on large mesh models downloaded from the net are extremely encouraging.