PhD defense : Segmentation of Deforming Meshes and Its Application to Similarity Measurement

PhD defense : Guoliang LUO

Team : IGG

Title : Segmentation of Deforming Meshes and Its Application to Similarity Measurement

Abstract : With an abundance of animation techniques available today, deforming meshes have become a subject of various data processing techniques in Computer Graphics community, e.g., shape retrieval and action recognition. Such tasks require efficient representations of animated meshes such as segmentation. Several spatial segmentation methods based on the movements of each vertex, or each triangle, have been presented in existing works that partition a given deforming mesh into rigid components. However, to the best of our knowledge, the segmentation of deforming meshes by taking into account of temporal deformation coherency has not been studied before. In this thesis, our contributions are the new segmentation techniques that compute the temporal and spatio-temporal segmentation for deforming meshes. In the first contribution, our temporal segmentation by minimizing the sum of within-segment dissimilarity can produce consistent temporal segmentation on different deforming meshes exhibiting similar motion, despite their shape differences. For the second contribution, we have developed a spatio-temporal segmentation method by investigating both the temporal and spatial deformation coherency of triangles. By representing the segmentation results into evolving graphs, we further propose a motion similarity measurement method by using a sequence alignment method. The experiment results show that our similarity measurement method successfully reflects human perception on the motion similarities of deforming meshes.

The jury is comprised of Hyewon Seo (University of Strasbourg), Frederic Cordier (Haute Alsace University), Stefanie Hahmann (Grenoble Institute of Technology), Mohamed Daoudi (Institut Mines-Télcom/Télécom Lille1), Atilla Baskurt (INSA Lyon) et Yong Xu (South China University of Technology).

The presentation will take place on Tuesday 4th november 2014 at 2.00pm in room A301 of the Pôle API building in Illkirch.