Motion and shape recovery from image frame sequences.
Date of Issue1998
School of Electrical and Electronic Engineering
The thesis focuses mainly on motion and structure recovery of an object from a sequence of image frames. The author's approach is different from the eight-point algorithm or other conventional non-linear optimization techniques. Full perspective projection is assumed in all the work. However it is shown that the methodologies proposed do encompass the weak perspective/orthographic projection model as well. Much of the thesis concentrates on finding what the minimum number of data points are needed for a unique solution under different number of frames and constraints.
DRNTU::Engineering::Computer science and engineering::Computing methodologies::Image processing and computer vision