Image Understanding in Unstructured Environment

In the development of autonomous sensory controlled systems, image understanding of sensory data is a difficult but important topic. Due to the unpredictable and uncertain nature of the environment, current image processing and computer vision approaches are not adequate to provide the capabilities needed by the systems. Thus, new approaches are required in the overall system design, including sophisticated reasoning processes, uncertainty management and adaptable architectures. This general issue is addressed by Thomas M Strat and Grahame B Smith. Lashon B Booker discusses the Bayesian approach in plausible reasoning for classification of complex ship images based on incomplete and uncertain evidence. Dynamic scene analysis is treated by Seetharaman Gunasekaran and Tzay Y Young. A spherical perspective approach is introduced to overcome some limitations of the current vision systems by Michael Penna and Su-shing Chen. Finally, Markov image models and their pixel-level approaches are extended to global approaches, through Dempster-Shafer and other techniques, by Mingchuan Zhang and Su-shing Chen.