H.-L. Chou (Taiwan)
Image-based rendering, projection model, shape from silhouette, color consistency check, image warping
Image-based rendering has become an active research area recently. It provides high quality rendered images without constructing detailed 3D object shape. The rendered images can be used in applications like products survey tools for E-commerce, computer games, digital museum browsing, and et cetera. In this paper, we propose a layered depth images generation and rendering method. The proposed method combines the silhouette and surface reflectance information to construct the shape of the object and then converts it to layered depth image format. The advantage of incorporating the color consistency check process into silhouette based reconstruction method is that we can use color information to carve out the resulting convex objects of silhouette based reconstruction method and produce the cavernous shape of cavity surface on the real object. After obtaining the layered depth images, we propose an algorithm which reduces the computation time required for rendering the layered depth images by reusing the previous computation and projection decomposition. Several experiments are conducted to demonstrate the performance of the presented method.
Important Links:
Go Back
