Buffered Adaptive Filters for Video Coding Bit Overhead
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current video coding systems face challenges in efficiently managing adaptive filters for motion prediction, leading to high bit costs and complexity in signaling filter coefficients, especially when using adaptive interpolation filters in block-based video coding.
Innovation Solution
The implementation of a Buffered Adaptive Filter (BAF) scheme, where adaptive filters from previous video frames are stored in a buffer and reused for future frames, reducing bit overhead by signaling only the filter index and using pixel and coefficient symmetries to minimize the number of unique coefficients needed.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If adaptive filters are used for motion prediction in video coding, then motion prediction accuracy is improved, but bit cost and signaling complexity increase
Solution Approach 1:
The patent applies preliminary action by pre-computing and storing adaptive filter coefficients in a buffer during encoding, so that when decoding, the system can retrieve and reuse these pre-computed filters without recalculating them. This reduces the bit cost of transmitting filter coefficients while maintaining motion prediction accuracy, as the filters are prepared in advance and reused across multiple frames.
Solution Approach 2:
The patent implements copying by storing adaptive filter coefficients in a buffer and reusing them for subsequent frames. Instead of transmitting new filter coefficients for each frame, the system copies and reuses previously computed filters, significantly reducing the bit overhead while preserving prediction accuracy through filter reuse across temporal references.
2Measurement precision
If adaptive filters are used for motion prediction in video coding, then motion prediction accuracy is improved, but signaling complexity increases
Solution Approach 1:
The patent applies preliminary action by pre-computing adaptive filter coefficients during encoding and storing them in a buffer. This preliminary computation eliminates the need for complex real-time signaling of filter coefficients during decoding, reducing signaling complexity while maintaining high motion prediction accuracy through reuse of pre-computed filters.
Solution Approach 2:
The patent implements copying by storing filter coefficients in a buffer and reusing them across frames. This approach simplifies signaling complexity by replacing complex real-time filter coefficient transmission with simple filter index retrieval from the buffer, while preserving prediction accuracy through filter reuse.
3Adaptability or versatility
If multiple adaptive filters are maintained for different reference pictures, then adaptability is improved, but buffer management complexity increases
Solution Approach 1:
The patent applies segmentation by organizing filter buffers separately for different reference pictures (e.g., refPic0 and refPic1). This segmentation allows independent management of filters for each reference picture, improving adaptability by enabling selective filter usage for different temporal references while reducing overall buffer management complexity through structured organization rather than a single large unified buffer.
Data Source
AI summary
A buffer is established in a video encoder and/or a video decoder. An adaptive filter is buffered in the established buffer. An input video signal is coded using the buffered filter. The filter buffer may be managed, e.g., as to buffer size or capacity, filter placement within the buffer, and buffered default filters. Adaptive filters may be signaled and buffered for multiple reference pictures.