Video Decoding Intra Prediction Quad Tree Optimization
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current video encoding and decoding methods face inefficiencies in calculating Rate-Distortion (RD) costs, particularly when the first-place inter 2N×2N mode is determined as the optimal prediction mode, leading to high time complexity and unnecessary encoding processes.
Innovation Solution
A method and apparatus for video decoding that reconstructs intra-prediction mode information and transform block coefficients in a quad tree structure, optimizing the decoding process by determining whether a block is divided into smaller blocks and only performing inverse transformations when necessary, thereby reducing calculation complexity and encoding time.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If RD costs are calculated for all PUs in a single CU to determine optimal mode, then encoding accuracy is improved, but time complexity increases significantly
Solution Approach 1:
The patent segments the CU into multiple PUs and processes them in a specific sequence (inter 2N×2N, inter 2N×N, inter N×2N, inter N×N, intra 2N×2N, intra N×N). By dividing the encoding process into discrete PU segments with a predetermined order, the system can terminate early when the first-place inter 2N×2N mode is optimal, avoiding unnecessary calculations for subsequent PUs while maintaining encoding accuracy for the selected mode.
Solution Approach 2:
The patent performs preliminary actions by establishing a fixed prediction sequence before encoding begins. The encoder determines the optimal PU type according to the predetermined sequence, and if the first inter 2N×2N mode achieves the best RD cost, the encoding process terminates immediately. This preliminary structuring of the encoding sequence allows early termination without compromising the accuracy of the optimal mode selection.
2Manufacturing precision
If all PUs are encoded to ensure optimal mode selection, then video quality is improved, but encoding speed decreases
Solution Approach 1:
The patent introduces dynamic termination into the encoding process. Instead of statically encoding all PUs regardless of outcome, the system dynamically evaluates RD costs in the predetermined sequence and terminates encoding as soon as the optimal mode is identified (when first-place inter 2N×2N mode is determined). This dynamic approach maintains video quality by ensuring optimal mode selection while significantly improving encoding speed through early termination.
Solution Approach 2:
The patent implements skipping by allowing the encoder to rush through the encoding process when the first inter 2N×2N mode proves optimal. Rather than methodically encoding all subsequent PUs, the system skips the remaining encoding steps entirely, achieving both high video quality through optimal mode selection and high encoding speed by eliminating unnecessary processing steps.
3Device complexity
If transform blocks are divided in quad tree structure and inverse transformation is performed only when necessary, then calculation complexity is reduced, but decoding accuracy may be compromised
Solution Approach 1:
The patent applies partial action by performing inverse transformation only on transform blocks that contain non-zero coefficients, rather than processing all blocks uniformly. The quad tree structure allows the decoder to identify and process only the necessary sub-blocks, reducing calculation complexity while maintaining decoding accuracy for the blocks that actually contribute to the reconstructed image.
Solution Approach 2:
The patent implements local quality by applying different processing strategies to different regions of the transform block structure. Blocks with non-zero coefficients receive full inverse transformation processing, while blocks with zero coefficients are skipped. This localized approach to processing maintains high decoding accuracy where needed while reducing overall calculation complexity by avoiding unnecessary processing in regions that don't contribute to image quality.
Data Source
AI summary
A video decoding apparatus using an intra-prediction, performs a method including: reconstructing information on an intra-prediction mode of a current block to be decoded from a bitstream, wherein the current block is a square block; reconstructing transform block information from the bitstream and reconstructing transform coefficients corresponding to each of one or more transform blocks divided in a quad tree structure from the current block; and reconstructing the current block based on the information on the intra-prediction mode and the reconstructed transform coefficients of the transform blocks.