Abstract:
|
Current literature features many approaches to statistical shape analysis, differing in representations, metrics and/or methods for shape alignment. Landmark-based methods ignore remaining outline information. More recently, elastic shape analysis attempts to fix this by using a special functional representation of the parametrically-defined outline to perform shape registration and further statistical analyses. However, lack of landmark identification can lead to unnatural alignment, particularly in biological and medical applications, where certain features are key to shape structure, comparison, and modeling. This work defines a joint landmark-constrained elastic shape analysis framework. Landmarks serve as constraints in the full shape analysis process, helping to disambiguate shape alignment when fully automatic elastic shape methods produce unsatisfactory solutions. We provide standard statistical tools on the landmark-constrained shape space including mean and covariance calculation, classification, clustering, and tangent PCA. We demonstrate the framework on shapes from the MPEG-7 dataset, and two real data examples: mice T2 vertebrae and Hawaiian Drosophila fly wings.
|