Correcting methods for efficient multiview video coding and improved display quality

点击翻译按钮获取中文摘要

Multiview video involves capturing 3D information from a scene, allowing exciting applications such as 3D TV and free viewpoint TV, while adding another dimension to surveillance and inspection by greatly increasing the accuracy of object recognition and tracking. Calibrating a large number of cameras is difficult, resulting in significant inconsistencies in brightness, colour, contrast and focus between the different views. These inconsistencies pose a number of problems in multiview systems, ranging from unpleasant visible mismatches for the viewers as they switch between different viewpoints to decreased efficiency in compression, rendering and tracking. The objective of our research is to develop digital pre-processing methods for correcting inconsistencies in multiview video sets, thus improving video coding performance and 3D display quality. Specifics include: