Adaptive Intra-refreshing Video Coding Units
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Video coding technologies face challenges in reducing I-pulsing artifacts, which occur due to data dependency in inter-predicted pictures, making them less resilient to data loss and susceptible to visually perceptible differences between intra-coded and inter-coded pictures.
Innovation Solution
The solution involves an adaptive approach that biases the decision between intra-prediction and inter-prediction for coding units in inter-coded pictures, favoring intra-prediction when pictures are temporally adjacent to intra-coded pictures, using a selection function that considers temporal distance, prediction errors, bit rate, and quantization factors to reduce I-pulsing artifacts.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Loss of energy
If inter-prediction is used for coding units in inter-coded pictures, then bit rate efficiency is improved, but I-pulsing artifacts increase and resilience to data loss deteriorates
Solution Approach 1:
The picture is divided into coding units that can be independently predicted using either intra-prediction or inter-prediction. The adapter selectively applies different prediction methods to different coding units based on their temporal distance from I-pictures, allowing some units to use efficient inter-prediction while others use robust intra-prediction, thus resolving the contradiction between bit rate efficiency and resilience to data loss
Solution Approach 2:
Different coding units within the same picture are assigned different prediction qualities based on their characteristics. Coding units closer to I-pictures use intra-prediction for robustness, while those farther away use inter-prediction for efficiency. This local differentiation allows the system to optimize both bit rate efficiency and resilience to data loss simultaneously
2Loss of energy
If inter-prediction is used for coding units, then bit rate efficiency is improved, but I-pulsing artifacts increase
Solution Approach 1:
The picture is segmented into coding units with different temporal distances from I-pictures. The adapter applies intra-prediction to units closer to I-pictures (reducing I-pulsing artifacts) and inter-prediction to units farther away (maintaining bit rate efficiency), thus resolving the contradiction between bit rate efficiency and I-pulsing artifacts
Solution Approach 2:
Different prediction methods are applied locally to different coding units based on their temporal distance from I-pictures. This local quality differentiation allows the system to minimize I-pulsing artifacts in regions prone to them while maintaining overall bit rate efficiency through inter-prediction in other regions
3Object-generated harmful factors
If intra-prediction is used for coding units temporally adjacent to I-pictures, then I-pulsing artifacts are reduced, but bit rate efficiency decreases
Solution Approach 1:
The picture is divided into coding units based on temporal distance from I-pictures. The adapter applies intra-prediction only to units closest to I-pictures (where I-pulsing artifacts are most problematic) and inter-prediction to other units, thus reducing I-pulsing artifacts while minimizing the bit rate cost
Solution Approach 2:
Intra-prediction is applied locally only where necessary (to coding units temporally adjacent to I-pictures) to reduce I-pulsing artifacts, while inter-prediction is used in other regions to maintain bit rate efficiency. This selective local application resolves the contradiction between artifact reduction and efficiency
Data Source
AI summary
The present disclosure relates to the use of adaptive intra-refreshing of video coding units to improve video perceptive quality by reducing artifacts such as I-pulsing. A picture to be encoded as an inter-coded picture is received from a video stream. A coding unit of the picture is encoded using an inter-predicted mode or an intra-predicted mode, where the mode is selected according to a selection function applied to the coding unit. The selection function is biased based at least in part on a temporal distance between the picture and a nearest intra-coded picture in the video stream. In various embodiments, bit rate, distortion from prediction errors, quantization factors and differences between pictures, content information, hypothetical reference decoder buffer information, group of pictures length, position of the coding unit, and/or other information may be factors employed in the selection function.