Dissertation: Image Sequence Processing using 3-D Segmentation - In this research, a sequence of images is viewed as a 3-D (spatiotemporal) volume. The important objects in a scene can therefore be associated with spatiotemporal regions in this volume. Using a Gibbs-Markov image model, these regions can be found (the sequence can be segmented) with great accuracy. The sequence can then be efficiently encoded in terms of region boundary surfaces and interior descriptions. This representation is also useful as a basis for image sequence processing techniques, such as noise reduction and image sharpening.