2DHV Interpolation Filters for Video Motion Compensation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current motion compensation techniques in video coding, such as those used in HEVC, rely on 1D interpolation filters applied separately in horizontal and vertical directions, which may not adequately address fractional pixel motion, leading to suboptimal compression efficiency.
Innovation Solution
Implementing 2D horizontal and vertical (2DHV) interpolation filters instead of traditional 1D filters, allowing for more flexible and adaptive interpolation based on block prediction types and quantization parameters, and using a CU-level flag to switch between different filter types for improved fractional-pel interpolation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Loss of energy
If 1D interpolation filters are used for motion compensation, then the device complexity is reduced, but the compression efficiency deteriorates due to inadequate handling of fractional pixel motion
Solution Approach 1:
The patent transitions from 1D interpolation filters to 2D interpolation filters for motion compensation. This dimensional change allows the filter to simultaneously consider horizontal and vertical pixel relationships, providing more accurate fractional pixel motion estimation and improving compression efficiency without proportionally increasing complexity.
Solution Approach 2:
The patent applies different interpolation filter types (e.g., 2DHV filters) selectively based on block characteristics such as prediction type and quantization parameter. This local differentiation optimizes interpolation quality for specific block regions while avoiding unnecessary complexity in other regions, resolving the contradiction between compression efficiency and device complexity.
2Measurement precision
If 2D interpolation filters are implemented for fractional pixel motion, then the interpolation precision is improved, but the computational complexity increases
Solution Approach 1:
The patent changes the parameters of the interpolation process by introducing 2D filtering with specific filter coefficients and structures. This allows achieving higher interpolation precision for fractional pixel motion while controlling computational complexity through optimized filter designs and selective application based on block characteristics.
Solution Approach 2:
Different 2D interpolation filter types are applied selectively to different block types (uni-predicted vs. bi-predicted blocks) based on quantization parameters. This localized approach ensures high precision where needed while reducing computational complexity in regions where simpler filters suffice.
3Loss of energy
If different interpolation filters are applied to uni-predicted and bi-predicted blocks, then the compression efficiency is improved, but the device complexity increases
Solution Approach 1:
The patent applies different interpolation filter types selectively to uni-predicted and bi-predicted blocks based on their specific characteristics and quantization parameters. This local differentiation improves compression efficiency by optimizing interpolation for each block type while the conditional logic for filter selection keeps the increase in device complexity manageable.
Solution Approach 2:
The patent introduces dynamic filter selection based on prediction type and quantization parameter conditions. This dynamic approach allows the system to adaptively choose the appropriate interpolation filter for each block, improving overall compression efficiency while the structured decision logic prevents excessive complexity increase.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
A method and user equipment for decoding a bitstream of video. The user equipment includes a transceiver and a decoder. The transceiver is configured to receive an encoded bitstream of video including a pixel block. The decoder includes processing circuitry configured to select, from among a plurality of vertical interpolation filters, one of a vertical one-dimensional filter and a vertical two-dimensional filter. The processing circuitry is also configured to select, from among a plurality of horizontal interpolation filters, one of a horizontal one-dimensional filter and a horizontal two-dimensional filter. The processing circuitry is also configured to use the selected vertical and horizontal interpolation filters to generate at least one pixel value by interpolating pixels of the pixel block. The processing circuitry is also configured to perform prediction decoding using the at least one pixel value to restore the video.