I. Introduction
A wide variety of future interactive multimedia applications, including free-viewpoint video or free-viewpoint television, 3-D television, immersive teleconference, and surveillance, are emerging. The existing and emerging technologies for modeling, coding, and rendering of real-world scenes for these applications have been reviewed at the convergence point of computer graphics, computer vision, and classical media [1]. It has been widely recognized that multiview video coding (MVC) is one of the key technologies for interactive multimedia applications [2]. Because of the huge data volume, efficient compression becomes more important in storage and delivery of multiview video than any other data. In most cases, multiview video is captured by multiple cameras from different angles and locations at the same time instant, and significant correlations may exist between the adjacent views.