Temporal Motion Vector Signaling Overhead Reduction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In video compression systems, particularly in High Efficiency Video Coding (HEVC), the overhead associated with signaling flags and indices for determining the reference list for temporal motion vectors is significant, as it requires encoding and decoding of flags like collocated_from_l0_flag and collocated_ref_idx, which can increase bitstream size and computational complexity.
Innovation Solution
A method is introduced to determine the value of a flag that controls the use of temporal motion vectors, where if the flag is set to a specific value, conditions are evaluated to decide whether to encode or decode the collocated reference picture list and position information, and if not, these elements are omitted from the bitstream, reducing overhead by not encoding or decoding the collocated_from_l0_flag and collocated_ref_idx when temporal motion vector predictors are not used.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If flags and indices for determining the reference list are signaled in the bitstream, then the temporal motion vector prediction can be accurately controlled, but the bitstream overhead increases
Solution Approach 1:
The patent extracts and removes unnecessary signaling elements (collocated_from_l0_flag and collocated_ref_idx) from the bitstream when they are not needed for temporal motion vector prediction. This is achieved by checking whether temporal MVP is actually used, and if not, omitting these flags and indices entirely, thereby reducing bitstream overhead while maintaining prediction accuracy when needed
Solution Approach 2:
The patent implements dynamic signaling where the inclusion of collocated reference picture information in the bitstream is conditional rather than fixed. The signaling is activated only when temporal motion vector prediction is used, and deactivated otherwise. This dynamic approach adapts the bitstream content to the actual prediction mode, reducing overhead in cases where temporal MVP is not applied
2Measurement precision
If flags and indices for determining the reference list are decoded, then the reference picture selection can be precisely determined, but the computational complexity increases
Solution Approach 1:
The patent removes the decoding of collocated_from_l0_flag and collocated_ref_idx when temporal motion vector prediction is not used. By extracting these unnecessary decoding operations based on the prediction mode, the system reduces computational complexity in the decoder while preserving precise reference picture selection capability when temporal MVP is activated
Solution Approach 2:
The patent implements dynamic decoding behavior where the decoder conditionally processes reference picture selection flags and indices based on whether temporal MVP is enabled. This dynamic approach reduces the number of decoding operations and computational steps required, lowering overall decoding complexity while maintaining precision when needed
3Productivity
If temporal motion vector predictors are not used, then the compression efficiency can be improved by omitting unnecessary information, but the adaptability to different prediction modes is reduced
Solution Approach 1:
The patent employs dynamic signaling that adapts to the actual prediction mode being used. When temporal MVP is not used, the collocated reference picture information is omitted from the bitstream, improving compression efficiency. When temporal MVP is used, the information is included, maintaining prediction flexibility. This dynamic adaptation resolves the contradiction by making the system efficient in non-temporal-MVP cases while remaining flexible when temporal prediction is needed
Solution Approach 2:
The patent changes the signaling parameters (presence or absence of flags and indices) based on the prediction mode. By conditionally including or excluding collocated reference picture information according to whether temporal MVP is used, the system optimizes compression efficiency for each specific case while preserving the ability to adapt to different prediction modes through parameter adjustment
Data Source
AI summary
A method determines a value for a first flag that controls whether a temporal motion vector is used for a slice. If the value of the first flag is a first value, the method performs: evaluating a first condition to determine whether to encode or decode a second flag; if the first condition indicates the second flag should be used, encoding or decoding the second flag in encoding or decoding the slice; evaluating a second condition to determine whether to encode or decode position information that indicates a position that contains the collocated reference picture; and if the second condition indicates the position information should be used, encoding or decoding the position information in encoding or decoding the slice. If the first flag is a second value, the method does not encode or decode the second flag and the position information.