Split Update Process for Motion Compensated Temporal Filtering
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current motion compensated temporal filtering (MCTF) in open loop scalable video coding requires additional residual buffers, leading to higher memory demands and complex buffer management, while removing the update process compromises compression efficiency and video quality.
Innovation Solution
The method splits the update process into multiple update processes using a single reference picture and employs a local update process that allows block-by-block execution, eliminating the need for residual buffers by using motion information derived from the prediction process.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If residual buffers are used to store intermediate decoded pictures in MCTF, then the prediction and update processes can be executed, but memory demands increase and buffer management becomes complex
Solution Approach 1:
The patent extracts and removes the update process from the conventional MCTF framework, eliminating the need for residual buffers that store intermediate decoded pictures. By taking out the update process entirely and relying solely on prediction processes, the system avoids the memory management complexity associated with residual buffers while maintaining essential MCTF functionality
Solution Approach 2:
The patent segments the conventional MCTF process into separate prediction processes, eliminating the integrated update process. This segmentation allows each prediction process to operate independently using only conventional picture buffers, removing the need for additional residual buffers and simplifying the overall buffer management structure
2Device complexity
If the update process is removed from MCTF, then buffer management is simplified, but compression efficiency deteriorates and video quality degrades
Solution Approach 1:
The patent changes the operational parameters of the prediction processes by configuring them to perform both prediction and update functions. By adjusting the parameters and control logic of the prediction processes, the system achieves the functionality of conventional MCTF (including update operations) while using only simplified buffer management structures
3Measurement precision
If two conventional picture buffers are used to store residual pictures, then buffer precision requirements are met, but memory requirements roughly double
Solution Approach 1:
The patent extracts the update process that requires high-precision residual buffers, eliminating the need for specialized residual picture storage. By removing this process, the system maintains necessary precision for prediction operations using only conventional picture buffers, avoiding the need to double memory allocation for residual picture storage
Solution Approach 2:
The patent makes conventional picture buffers universal by having them perform both prediction and update storage functions. Instead of dedicating separate high-precision residual buffers for update operations, the same conventional buffers serve multiple purposes, reducing total memory requirements while maintaining adequate precision for all operations
Data Source
AI summary
A method for temporal decomposition and reconstruction of an input video signal is disclosed. The method uses a prediction process and an update process in the framework of motion compensated temporal filtering (MCTF), the motion information used for the update process being derived from the motion information used for the prediction process, and the method employing a block based video codec composed of an encoder and a decoder. In response to the update and prediction processes on a picture being based on N reference pictures where N is greater than one, the update process is split into N update processes using a single picture as a reference. Each one of the split update processes is executed each time a picture that is needed as a reference by the process is received.