CMStatistics 2015: Start Registration
View Submission - CMStatistics
B1029
Title: Models and statistics for projective shape analysis Authors:  John Kent - University of Leeds (United Kingdom) [presenting]
Abstract: The projective shape of a geometric object is the information that is invariant under projective transformations. The main application is to camera images, where the choice of projective transformation, or pose, corresponds to the camera view of the object. The simplest example of a projective shape is the cross ratio for a set of four collinear points. The way in which measurement errors in a camera image of an object affect the observed projective shape depends on the pose of the object. Thus the statistical analyis for a collection of images involves an interplay between the underlying projective shapes and the estimated poses.