Video Encoder Interpolation Method for Filtering Accuracy
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current H.265 encoding and decoding techniques require improvements in processing efficiency, particularly in generating prediction images, as they often rely on fixed interpolation methods that do not account for the characteristics of blocks in images, leading to suboptimal memory usage and filtering accuracy.
Innovation Solution
An encoder and decoder that determine an interpolation method based on the characteristics of blocks, such as gradient directionality, to interpolate pixels outside a referable region, allowing for more accurate filtering without storing actual pixels outside the region, thereby reducing memory requirements and enhancing filtering accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If actual pixels outside the referable region are stored in memory for filtering, then filtering accuracy is improved, but memory capacity requirements increase
Solution Approach 1:
Instead of storing actual pixels outside the referable region, the patent creates interpolated copies of these pixels by extrapolating from pixels within the referable region. This copying approach provides sufficient pixel data for filtering operations without requiring additional memory capacity for storing actual external pixels.
Solution Approach 2:
The patent introduces an interpolation process as an intermediary between the referable region and the filtering operation. This intermediary generates the necessary pixel values through extrapolation methods (such as replication or mirroring of boundary pixels) without requiring direct access to or storage of actual pixels outside the referable region.
2Ease of operation
If a fixed interpolation method is used for all blocks, then processing simplicity is improved, but filtering accuracy deteriorates
Solution Approach 1:
The patent applies different interpolation methods to different blocks based on their local characteristics. By analyzing properties such as gradient directionality and block type (e.g., DC prediction blocks vs. non-DC blocks), the system selects appropriate interpolation strategies for each local region, thereby improving filtering accuracy while maintaining reasonable processing complexity.
Solution Approach 2:
The patent changes the interpolation parameters (method selection) based on block characteristics. By dynamically adjusting which interpolation technique is applied to each block according to its specific properties, the system achieves better filtering accuracy without requiring overly complex processing for all blocks uniformly.
Data Source
AI summary
An encoder includes processing circuitry and memory. Using the memory, the processing circuitry: encodes and reconstructs an image to generate a reconstructed image; determines, according to a characteristic of a block in the reconstructed image, an interpolation method for interpolating pixels located outside a referable region including the block; interpolates the pixels located outside the referable region, using the interpolation method determined; and applies a filter to the block using the pixels interpolated.