SVC Motion Vector Prediction Upsampling Interpolation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The H.264/AVC standard lacks efficient methods for scalable video coding, particularly in spatial scalability, where the accuracy of motion vectors (MV) for the BLSkip mode is low due to half-pixel accuracy and shared MVs across 4×4 blocks within an 8×8 block, affecting coding efficiency.
Innovation Solution
The method involves upsampling and interpolating the motion field from 8×8 to 4×4 blocks, using scalar median, vector median, and average methods to select a better predictor for the BLSkip mode by smoothing and interpolating motion vectors, and generating predictors from neighboring blocks.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If the base layer motion field is upsampled by multiplication with two to obtain enhancement layer motion vectors, then the device complexity is reduced and ease of operation is improved, but the motion vector accuracy deteriorates from 1/4 pixel to 1/2 pixel accuracy
Solution Approach 1:
The patent applies preliminary action by performing interpolation of the base layer motion field before upsampling. The motion compensation residual from the base layer is interpolated to sub-sample accuracy (1/4 or 1/8 pixel) prior to the upsampling operation, so that the enhancement layer motion vectors inherit this higher accuracy. This preliminary interpolation step ensures that the simplicity of multiplication-based upsampling is maintained while achieving superior motion vector precision.
2Device complexity
If four 4×4 blocks share the same upsampled motion vector from an 8×8 block, then the device complexity is reduced, but the manufacturing precision of motion compensation deteriorates
Solution Approach 1:
The patent applies segmentation by dividing the enhancement layer macroblock into four separate 4×4 blocks, each with its own independently interpolated motion vector. Instead of assigning a single upsampled motion vector to all four blocks, the method performs separate interpolation for each 4×4 block based on the base layer motion field and motion compensation residuals. This segmentation enables each block to have customized motion parameters, improving motion compensation precision while maintaining reasonable computational complexity.
3Loss of information
If motion vectors are not coded during decode process in MODE-8×8, then the loss of information is reduced and coding efficiency is improved, but the reliability of motion prediction deteriorates due to lower accuracy
Solution Approach 1:
The patent applies feedback by using the motion compensation residual from the base layer decoding process to improve the enhancement layer motion vector prediction. The residual signal, which contains information about prediction errors, is interpolated and used to refine the enhancement layer motion vectors. This feedback mechanism allows the system to achieve more accurate motion prediction without coding explicit motion vectors for the enhancement layer, thereby maintaining coding efficiency while improving prediction reliability.
Data Source
AI summary
A method for improving the performance of the BLSkip mode in SVC includes the steps of upsampling the motion field of the base layer, interpolating the motion vectors for the intra MBs, interpolating the 8×8 block motion field to a 4×4 block motion field, and generating a MV predictor for a 4×4 block in BLSkip mode using neighbor candidates.