This dissertation investigates the challenge of accurately estimating depth from a single camera's video feed. The author, Hon-tat Ho, addresses the limitations of monocular vision in robotics and computer vision by proposing an integrated approach that leverages motion and visual perception principles. The work utilizes image processing techniques to infer three-dimensional spatial information from a sequence of two-dimensional images, aiming to enhance the capabilities of mobile robots in complex environments.

This dissertation is a technical academic work, likely to be of interest to researchers and students in computer vision, robotics, and image processing. Its focus on an integrated approach suggests a comprehensive treatment of the subject matter, aiming to build upon existing knowledge in the field. The work is presented as a dissertation from The University of Hong Kong, indicating a rigorous academic standard and contribution to the field.