摘要:
An apparatus and method for video coding having intra-slice deblocking, intra-slice adaptive loop filter, and intra-slice adaptive offset are disclosed. In a video coding system, a picture is often divided into slices and each slice is independently processed so that errors or missing data from one slice cannot propagate to any other slice within the picture. In the recent high efficiency video coding (HEVC) development, deblock filtering (DF), adaptive loop filter (ALF) and adaptive offset (AO) may be applied to reconstructed slices. When the processing is applied across slice boundaries, it may rely on pixels outside the current slice and cause dependency of the current slice on other slice(s). Consequently, the DF/ALF/AO processing on a reconstructed slice may have to wait until the reconstructed slices that it is dependent from complete the DF/ALF/AO processing. To overcome the slice boundary issue, intra-slice DF/ALF/AO is developed that does not rely on any pixel outside the currently slice. An optional slice boundary filter may be applied to slice boundaries after all slices are processed by the intra-slice DF/ALF/AO.
摘要:
An apparatus and method for motion vector prediction for a current block in a picture are disclosed. In video coding systems, the spatial and temporal redundancy is exploited using spatial and temporal prediction to reduce the information to be transmitted. Motion Vector Prediction (MVP) has been used to further conserve the bitrate associated with motion vector. In conventional temporal MVP, the predictor is often based on a single candidate such as the co-located motion vector in the previous frame/picture. If the co-located motion vector in the previous frame/picture does not exist, the predictor for the current block is not available. A technique for improved MVP is disclosed where the MVP utilized multiple candidates based on co-located motion vectors from future and/or past reference pictures. The candidates are arranged according to priority order to provide better availability of MVP and also to provide more accurate prediction. Furthermore, the MVP technique disclosed can be operated in a closed-loop fashion so that no additional side information or minimum additional side information is required.
摘要:
A motion prediction method is provided. The method has the following steps of: determining a plurality of candidate units corresponding to a current unit of a current frame; obtaining a plurality of motion vectors of the candidate units; calculating a plurality of temporal scaling factors of the candidate units according to a plurality of temporal distances between a plurality of reference frames of the motion vectors and the current frame; scaling the motion vectors of the candidate units according to the temporal scaling factors to obtain a plurality of scaled motion vectors; and selecting a motion vector predictor for motion prediction of the current unit from the candidate units according to the scaled motion vectors.
摘要:
A method and apparatus for encoding a video sequence comprising a plurality of pictures are disclosed. In video coding systems, the temporal redundancy is exploited using motion compensated prediction. The video sequence is often organized into multiple GOP (group of pictures) where different types of GOP may be used. In conventional coding systems, IPPP and IBBP GOP structure is often used. In H.264/AVC and the emerging High Efficiency Video Coding (HEVC), hierarchical GOP structure, including hierarchical P GOP structure and hierarchical B GOP structure, has been introduced to allow temporal scalability. Furthermore, low-delay IBBB GOP structure has been also introduced, for low-delay application. In the present invention, a low-delay hierarchical B GOP structure is disclosed. The new structure uses low-delay B-pictures only so as to minimize the processing delay while the hierarchical structure provides the temporal scalability. The low-delay hierarchical B GOP structure has been shown to result in substantial improvement in coding efficiency.
摘要:
In a block-based motion compensated system, coding noises may arise due to lossy operations. Adaptive loop filter (ALF) is a technique used to improve the quality. A pixel adaptive (PA) ALF method was developed that uses the Sum-modified Laplacian Measure (SLM) to classify pixels and applies a respective ALF filter to each pixel according to the SLM. While the PA ALF achieves better performance over a conventional single filter (SF) based ALF, the PA ALF causes higher complexity and consumes more power due to the high complexity and per-pixel filter switching. Accordingly region based ALF scheme is disclosed which allows adaptive filter selection on a region by region basis and does not require per-pixel filter switching. In one embodiment according to the present invention, a picture is divided into M×N regions. The region based ALF also allows region merging for the M×N regions to improve performance. In another aspect of the present invention, the optimal ALF design also takes into consideration of the system complexity by using rate-distortion-complexity optimization (RDCO). The RDCO technique is also applied to sample adaptive offset (SAO) design.
摘要:
An apparatus and method for deriving a motion vector predictor (MVP) or a MVP candidate for a current block are disclosed. Embodiments according of the present invention receive a first motion vector associated a first reference picture in a first reference picture list and a second motion vector associated with a second reference picture in a second reference picture list of a spatially neighboring block. A MVP or at least one MVP candidate associated with a selected reference picture in a selected reference picture list for the current block is then determined based on the first reference picture, the second reference picture and the selected reference picture according to a pre-defined priority order. The MVP or MVP candidate is determined depending on whether the first reference picture is the same as the selected reference picture or whether the second reference picture is the same as the selected reference picture.
摘要:
A method of performing block matching on a video image of progressive data includes: receiving the video image; storing the video image to a first storage; loading a current macroblock pair and a current search window of the video image to a second storage and utilizing the current search window to perform block matching for the current macroblock pair; and loading a next macroblock pair and a plurality of macroblocks corresponding to a non-overlapped area of a next search window to the second storage to perform block matching for the next macroblock pair.
摘要:
A video encoder that utilizes adaptive loop filtering for coding video data includes: a prediction unit, for performing prediction techniques according to original video data and reconstructed video data to generate predicted samples; a reconstruction unit, for reconstructing the predicted samples to form the reconstructed video data; a reference picture buffer, for storing the reconstructed video data; a filter parameter estimator, for estimating filter parameters according to the original video data and the reconstructed video data; and an adaptive filter, for utilizing the stored filter parameters to perform filtering on the reconstructed video data.
摘要:
An apparatus and method for deriving a motion vector predictor are disclosed. A search set comprising of multiple (spatial, or temporal) search MVs with priority is determined, wherein the search MVs for multiple neighboring reference block or one or more co-located reference blocks arc configured into multiple search MV groups. In order to improve coding efficiency, embodiments according to the present invention, perform redundancy check every time after a search MV group is searched to determine whether an available search MV found. If an available search MV is found and the available search MV is not the same as a previously derived motion vector predictor (MVP), the available search MV is used as the MVP and the MVP derivation process terminates. Otherwise, the MVP derivation process moves to the next reference block. The search MV group can be configured to include different search MV(s) associated with reference blocks.
摘要:
Methods, encoders, and decoders with localized weighted prediction are disclosed. A decoding method includes decoding data for a current segment to generate decoded data including residuals and a weighted prediction parameter for the current segment. A weighted prediction for the current segment is generated based on the weighted prediction parameter. A predictor for the current segment is generated by intra/inter prediction. The weighted prediction and the predictor are combined to obtain a modified predictor, and the current segment is reconstructed according to the modified predictor and the residuals.