Automated feature detection in multidimensional images: a unified tensor approach
2001 (English)Licentiate thesis, comprehensive summary (Other academic)
Manual identification of structures and features in multidimensional images is at best time consuming and operator dependent. Feature identification need to be accurate, repeatable and quantitative. This thesis presents a unified approach for automatic feature detection in multidimensional scalar and vector fields. The basis for the feature detection is a tensor representation of multidimensional local neighborhoods, constructed from a filter response controlled linear combination of basis tensors. Using eigenvalue and eigenvector decomposition, local topology and local orientation can be estimated. With different filter sets the tensor representation can be used to find specific features in multidimensional images, such as planar structures in scalar fields or flow structures as vortices or parallel flow in vector fields.
The motivation for the unified approach for feature detection in both scalar and vector fields is to build a foundation to tackle the challenge of understanding the complex interaction between the cardiac walls and the blood flow in the human heart.
Place, publisher, year, edition, pages
Linköping: Linköpings universitet , 2001. , 42 p.
Linköping Studies in Science and Technology. Thesis, ISSN 0280-7971 ; 909
Medical and Health Sciences
IdentifiersURN: urn:nbn:se:liu:diva-33625Local ID: LiU-TEK-LIC-2001:46ISBN: 91-7373-131-5OAI: oai:DiVA.org:liu-33625DiVA: diva2:254448
2001-11-30, Terassen, Universitetssjukhuset, Linköping, 10:15 (Swedish)
Nordberg, Klas, PhD
List of papers