Video Prediction Update Switching for Faster Encoder-Decoder Pipelines
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing video coding technologies face challenges in efficiently processing the increasing amount of digital video data, particularly in terms of encoding and decoding efficiency and speed, without increasing processing time or circuit size.
Innovation Solution
The proposed encoder and decoder systems utilize a bi-directional optical flow (BDOF) process and intra prediction mixing process, which are mutually exclusive at specific stages of the encoding/decoding pipeline, to generate prediction images, thereby optimizing the encoding/decoding process and reducing processing time and circuit size.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If both BDOF process and intra prediction mixing process are applied at the same stage, then prediction accuracy is improved, but processing time and circuit size increase
Solution Approach 1:
The patent implements dynamic selection of prediction processes by introducing a flag signal that indicates whether the current block is a skip block. Based on this flag, the system dynamically switches between applying BDOF process, intra prediction mixing process, or neither, allowing the prediction mechanism to adapt its complexity to the specific block characteristics and motion patterns.
Solution Approach 2:
The patent changes the parameter of process application by using a flag signal to control whether BDOF and/or intra prediction mixing processes are applied. This parameter change allows the system to optimize between prediction accuracy and processing efficiency by selectively enabling or disabling specific prediction processes based on block type and motion characteristics.
2Measurement precision
If both BDOF process and intra prediction mixing process are applied at the same stage, then prediction accuracy is improved, but circuit size increases
Solution Approach 1:
The patent implements dynamic selection of prediction processes by introducing a flag signal that indicates whether the current block is a skip block. Based on this flag, the system dynamically switches between applying BDOF process, intra prediction mixing process, or neither, allowing the prediction mechanism to adapt its complexity to the specific block characteristics and motion patterns.
Solution Approach 2:
The patent changes the parameter of process application by using a flag signal to control whether BDOF and/or intra prediction mixing processes are applied. This parameter change allows the system to optimize between prediction accuracy and processing efficiency by selectively enabling or disabling specific prediction processes based on block type and motion characteristics.
3Measurement precision
If multiple prediction processes are applied sequentially, then prediction quality is improved, but the number of pipeline stages increases
Solution Approach 1:
The patent implements dynamic selection of prediction processes by introducing a flag signal that indicates whether the current block is a skip block. Based on this flag, the system dynamically switches between applying BDOF process, intra prediction mixing process, or neither, allowing the prediction mechanism to adapt its complexity to the specific block characteristics and motion patterns.
Solution Approach 2:
The patent changes the parameter of process application by using a flag signal to control whether BDOF and/or intra prediction mixing processes are applied. This parameter change allows the system to optimize between prediction accuracy and processing efficiency by selectively enabling or disabling specific prediction processes based on block type and motion characteristics.
Data Source
AI summary
An encoder: generates, in an inter prediction mode, a first prediction image of a current block to be processed, based on a derived motion vector; and generates a final prediction image of the current block by applying an update process to the first prediction image. Candidates for the update process include a first process and a second process. The first process is a BDOF process. The second process is a process of mixing the first prediction image with a second prediction image generated in intra prediction for the current block. In the applying of the update process, the first process and the second process are mutually exclusively applied.