Recent advances in the area of computer vision and 3D-acquisition techniques have led to rapidly growing amount of 3D-data. Furthermore, the World Wide Web is enabling access to 3D models constructed by people all over the world. Currently, several web sites allow free downloads for thousands of publicly available 3D models. 3D models will increasingly be an important part of Digital Libraries in the future.

These developments are changing the way we deal with the problem of 3D data generation. To date, the field of 3D modeling concentrated on how to model artificial representations of 3D entities from scratch or from a few already available objects. In the future, a central question will be how to 'recycle' existing 3D models, how to analyze them and how to adapt them to our own purposes. An important issue in this general context is how to search for 3D objects like we already search for text, images, audio, and video.

The goal of our project is to elaborate a set of algorithms and methods that will allow our system to compare 3D object in a way that is close to the human notion of resemblance. As a further step, these results will be used to automatically search in 3D Digital Libraries.