摘要:
A method and apparatus is disclosed herein for motion vector prediction and coding. In one embodiment, the method comprises: deriving N motion vector predictors for a first block that has N motion vectors corresponding to N lists of reference frames and a current frame, including constructing one of the N motion vector predictors when a second block that neighbors the first block and is used for prediction has at least one invalid motion vector, where N is an integer greater than 1; generating N differential motion vectors based on the N motion vectors and N motion vector predictors; and encoding the N differential motion vectors.
摘要:
A method for encoding a non-manifold polygonal mesh operates by converting an original (non-manifold) model to a manifold model, encoding the manifold model using a mesh compression technique, and clustering, or stitching together during a decompression process, vertices that were duplicated earlier to faithfully recover the original connectivity. By separating the connectivity from model geometry and properties, the method avoids encoding vertices (and properties bound to vertices) multiple times. This provides a reduction in the size of the bit-stream compared with encoding the model as a manifold.
摘要:
A data structure for representing a general n-dimensional polygonal mesh. The data structure includes a structure record and a data record for each three dimensional shape. The structural record contains polygonal model connectivity information and further includes a stitching record that defines corresponding polygonal (triangular) mesh edge pairs and a polygonal (triangular) tree record representing a polygon (triangle) tree. The stitching record includes a vertex tree and a set of jump edges. The data record includes at least three polygonal records, each corresponding to a polygon. Each polygonal record is associated with a face of said polygonal model and classifies its corresponding polygon as either a leaf polygon, a running polygon or a branching polygon. Polygonal shapes are encoded into the data structure by first building a spanning tree for the polygonal mesh. A set of cut edges are derived for the polygonal mesh. The stitching record is constructed for the set of cut edges. Then traversing the polygon tree, the data records are encoded. The encoded data is transmitted, the structure record being transmitted first, followed by the data record. The stitching tree and polygon tree are decoded as they are received and, after at least three polygons are received, received polygons may be displayed.
摘要:
A method and apparatus is disclosed herein for motion vector prediction and coding. In one embodiment, the method comprises: deriving N motion vector predictors for a first block that has N motion vectors corresponding to N lists of reference frames and a current frame, including constructing one of the N motion vector predictors when a second block that neighbors the first block and is used for prediction has at least one invalid motion vector, where N is an integer greater than 1; generating N differential motion vectors based on the N motion vectors and N motion vector predictors; and encoding the N differential motion vectors.