Rectangular Filter Codebook for Video Compression
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing video coding standards like H.264 and HEVC face challenges in achieving high compression efficiency for high-resolution video applications, leading to high bitrate and bandwidth requirements, and existing filtering methods, such as Quality Restoration (QR) filtering, do not fully address decoding complexity and compression efficiency.
Innovation Solution
A hybrid subset of rectangular or square filters, such as 11×9 or 9×9, are used for efficient filtering, with codebook search to minimize overhead, where computed coefficients or filter indices are encoded and sent to the decoder, allowing both luma and chroma signals to be filtered with different shapes and sizes, improving prediction and reconstruction signal efficiency.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Quantity of substance
If traditional video coding standards (H.264, HEVC) are used for high-resolution video, then video compression is achieved, but bitrate and bandwidth requirements become relatively high
Solution Approach 1:
The patent transforms the filtering approach by changing parameters from traditional square filters to rectangular filters with varying aspect ratios (e.g., 11×9, 9×11). This parameter change allows adaptive filtering that better matches the anisotropic characteristics of video content, improving compression efficiency by 0.5-1.5 dB PSNR while reducing bitrate requirements
Solution Approach 2:
The patent introduces dynamic filter selection where the encoder chooses from multiple rectangular filter shapes based on local content characteristics. The filter shape and strength are adapted dynamically for different regions (e.g., horizontal vs. vertical edges, smooth vs. textured areas), enabling optimal compression at reduced bitrate compared to static traditional filters
2Manufacturing precision
If advanced filtering methods are applied to improve video quality, then decoded video quality improves, but decoding complexity increases
Solution Approach 1:
The patent segments the filtering process into distinct rectangular filter operations that can be independently applied to different video regions. By dividing the complex filtering task into simpler rectangular convolution operations with predefined kernels, the decoder complexity is reduced while maintaining high video quality through region-specific filtering
Solution Approach 2:
The patent uses codebook-based filtering where pre-computed filter kernels are stored in a codebook and referenced during decoding. Instead of computing complex filters in real-time, the decoder copies and applies pre-optimized rectangular filter kernels from the codebook, significantly reducing decoding complexity while preserving video quality
3Productivity
If rectangular filters with different shapes and sizes are used for luma and chroma signals, then filtering efficiency improves, but overhead increases
Solution Approach 1:
The patent creates a universal rectangular filter framework that handles both luma and chroma signals using the same filter shapes (e.g., 11×9, 9×11). This multi-functional approach allows a single set of rectangular filter kernels to serve multiple signal types, improving filtering efficiency while reducing overhead compared to having separate filter sets for each signal type
Data Source
AI summary
Techniques related to quality restoration filtering for video coding are described.