|M.Sc Thesis||Department of Electrical Engineering|
|Supervisor:||Prof. Tal Ayellet|
|Full Thesis text|
This thesis investigates a novel approach to performing partial matching. Given two meshes MA and MB, our goal is to find the best match to MA within MB. Our observation is that though isolated feature points often do not suffice, their aggregation provides adequate information regarding similarity. Hence, the key idea of our approach is to integrate feature-point similarity and segment similarity. Specifically, we introduce a probabilistic framework in which the segmentation and the correspondences of neighboring feature points allow us to enhance or moderate our certainty of the feature-point similarity. We show that this scheme manages to detect partial similarity automatically for a wide class of objects. Partial matching is important in a variety of geometry processing applications, including self-similarity, registration, retrieval, modeling etc. We demonstrate the utility of our algorithm in two applications: self similarity and partial matching in archaeology. Self-similarity is used in shape analysis, modeling, design, detection of symmetries, and editing of 3D objects. In archaeology, after finding a new artifact, the major task of the archaeologist is to locate it in time and space. Being able to perform partial matching is vital to this task because often only small fragments of the artifacts are found. An automatic matching can serve as a worthy alternative to the expensive and time-consuming manual procedure that is used today.