Audio-visual experiences of modeled spaces play a key role when studying the human perception of the built environment, and the results should give us an intuitive sense of the real situation. Acoustic models need a low number of polygons in order to be simulated efficiently. It is at the same time important to retain acoustic properties while simplification. But the reconstruction of existing architecture is usually achieved by photogrammetry, which results in noisy and geometrically complex 3D meshes. They must be post-processed to become suitable for simulation.

In this thesis, a mesh-filtering algorithm is analyzed that follows a feature-preserving approach. This is expected to be an important step before the mesh’s number of polygons can be reduced. The algorithm has been implemented in Matlab. Two models serve as test scenarios, i.e. a photogrammetry mesh of the Eurogress hall in Aachen and a generic simple room. To evaluate the algorithm’s performance, first, the change of geometrical shapes is inspected by means of the number of iterations and selected filter weights. Secondly, geometrical acoustic simulation is performed to analyze the filter’s impact on the acoustic simulation result.

Melden Sie sich hier an um Einladungen zu den Kolloquium-Vorträgen per E-Mail zu erhalten.
Register here to receive the invitations to colloquium talks via e-mail.