Motion Projection Error Concealment for Block-Based Video
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing error concealment algorithms for block-based video decoding fail to effectively handle entire frame losses, especially in low-bit-rate applications like 3G wireless networks, and often result in low PSNR values and noticeable error propagation due to chaotic motion.
Innovation Solution
A block-based error concealment method that projects motion vectors from correctly received blocks to lost blocks, with a two-stage process: motion-projection and post-processing, where motion vectors are reassigned based on overlap and refined using median filtering to minimize discontinuities.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of manufacture
If Frame Copy is used for error concealment, then implementation is simple, but PSNR value is low and error propagation is noticeable for video sequences with active motion
Solution Approach 1:
The patent segments the video frame into multiple blocks and processes each block independently using motion vector projection. Instead of applying a global concealment method, the algorithm divides the frame into 16x16 macroblocks and further into 4x4 blocks, allowing localized motion compensation that adapts to different motion regions while maintaining overall simplicity.
Solution Approach 2:
The patent introduces dynamic motion vector projection that adapts to local motion characteristics. By projecting motion vectors from reference frames and selecting the best matching vectors for each block based on overlap criteria, the method dynamically adjusts to different motion patterns in different regions, improving concealment quality for active motion sequences.
2Reliability
If Motion Copy is used for error concealment, then it handles homogeneous motion fields well, but it generates concealment artifacts when there is abundant chaotic motion
Solution Approach 1:
The patent applies local quality by treating different blocks with different motion characteristics independently. Each block's motion vector is projected and selected based on its specific overlap criteria, allowing regions with homogeneous motion to use Motion Copy-like behavior while regions with chaotic motion receive localized adjustments, preventing artifact propagation.
Solution Approach 2:
The patent changes motion vector parameters dynamically by projecting multiple candidate motion vectors and selecting the best match for each block based on overlap metrics. This parameter selection approach allows the system to adapt to different motion patterns, using simple copied parameters for homogeneous regions and adjusted parameters for chaotic motion regions.
3Manufacturing precision
If pixel-level error concealment algorithms are used, then motion vectors are estimated for each pixel, but substantial additional computation is required and existing data structures must be changed
Solution Approach 1:
The patent segments the pixel-level motion estimation problem into block-level processing. Instead of estimating motion vectors for each individual pixel, the algorithm works with 4x4 blocks within 16x16 macroblocks, significantly reducing computational complexity while maintaining adequate precision for error concealment applications.
Solution Approach 2:
The patent uses motion vector copying from reference frames as the basis for error concealment. By projecting and selecting motion vectors from already-decoded reference frames rather than performing new motion estimation, the algorithm avoids substantial additional computation while still providing improved concealment quality over simple Frame Copy.
4Device complexity
If block-level error concealment is used, then computational overhead is reduced, but no solution is provided for blocks that fail consistency checks
Solution Approach 1:
The patent performs preliminary motion vector projection and overlap assessment for all blocks before final selection. By pre-calculating projection positions and overlap metrics, the algorithm identifies suitable candidate blocks in advance, ensuring that even blocks with challenging motion characteristics have viable concealment options available.
Solution Approach 2:
The patent incorporates feedback mechanisms where the overlap criteria and projection results inform the selection process. Blocks that initially fail consistency checks can be re-evaluated using feedback from neighboring block selections and projection outcomes, ensuring reliable concealment is achieved for all blocks regardless of initial consistency check results.
Data Source
AI summary
There are provided a method and apparatus for block-based error concealment. The apparatus includes an error concealment module for concealing a current block in a lost picture of a video sequence by projecting positions of at least one block in a correctly received picture of the video sequence to a position of the current block in the lost picture using motion vectors of the at least one block in the correctly received picture so as to obtain a resultant projection, and selecting one of the at least one block in the correctly received picture based upon an amount of overlap in the resultant projection. The current block in the lost picture is then concealed using a motion vector of the selected one of the at least one block in the correctly received picture.