Video Coding Unit Partitioning for Inter Prediction Accuracy
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current video compression technologies, such as H.264/AVC, face limitations in achieving high-quality, high-performance compression, particularly in the inter prediction process from a reference frame, which affects coding efficiency and image quality.
Innovation Solution
The apparatus and method enhance video encoding and decoding by partitioning a coding unit (CU) into prediction units (PUs) of sizes 2N×hN or hN×2N, using a predictor to generate predicted blocks through intra or inter prediction, and a bitstream generator to encode these predictions, along with a transformer and quantizer to optimize frequency domain transformation and quantization.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If a coding unit (CU) is partitioned into prediction units (PUs) with sizes 2N×hN or hN×2N, then inter prediction performance is improved, but device complexity increases
Solution Approach 1:
The coding unit (CU) is divided into multiple prediction units (PUs) with different size configurations (2N×hN or hN×2N) to enable flexible partitioning. This segmentation allows the predictor to process smaller regions with finer granularity, improving inter prediction accuracy by better matching motion characteristics in different areas of the block.
Solution Approach 2:
The patent introduces dynamic PU mode selection where the encoder can adaptively choose between different PU partitioning configurations (2N×hN, hN×2N, or other modes) based on the specific content characteristics and motion patterns. This dynamic adaptation allows the system to optimize prediction performance for each particular coding situation without being locked into a fixed partitioning scheme.
2Productivity
If multiple PU modes are supported for flexible partitioning, then coding efficiency is improved, but bitstream overhead increases
Solution Approach 1:
The patent employs parameter change techniques where instead of transmitting complete PU mode information for every block, the system uses flags and indicators to signal PU mode decisions. By changing the representation parameters (using compact flag-based syntax rather than full mode enumeration), the system reduces the number of bits needed to convey PU partitioning information while maintaining the flexibility to support multiple PU modes.
Solution Approach 2:
The encoder performs rate-distortion optimization by evaluating different PU modes and selecting the one that provides the best compression efficiency. This feedback mechanism allows the system to adaptively choose PU modes that balance coding efficiency with bitstream complexity, avoiding modes that would increase overhead without providing significant quality improvement.
Data Source
AI summary
An apparatus for encoding images includes: a predictor to determine a prediction unit (PU) mode among a plurality of PU modes indicating types of partition of a current coding unit (CU) into one or more prediction units (PUs), and generate a predicted block of the current CU by performing an intra prediction or an inter prediction for each PU of the determined PU mode, wherein a size of the current CU is 2N×N, and the plurality of PU modes includes 2N×hN or hN×N PU mode in which the current CU is partitioned into PUs respectively having a size of 2N×hN or hN×N, h being a fractional number smaller than 1; a subtractor to subtract the predicted block from the current CU to generate a residual block; a transformer to transform the residual block into a frequency domain to generate a frequency one or more transform blocks.