Video Encoder Flicker Reduction via Bidirectional Prediction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing moving image encoding techniques face challenges in maintaining high encoding efficiency while reducing intra flicker, especially when using Closed-GOP formats, and existing methods are not effective in all applications, including those requiring independent GOP handling and motion prediction reduction.
Innovation Solution
A moving image encoding device that acquires a succeeding frame when encoding an inter picture, determines if the flicker criterion is met based on frame characteristics, and generates a provisional prediction image using a reference and succeeding frame for inter frame prediction, allowing for bidirectional prediction to reduce flicker.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If intra frame prediction is used for encoding, then encoding independence and random access capability are improved, but flicker occurs in the decoded moving image
Solution Approach 1:
The patent applies preliminary action by generating a provisional prediction image using bidirectional prediction before final encoding. This preliminary prediction image is created by combining forward prediction from a reference frame and backward prediction from a succeeding frame, allowing the system to anticipate and compensate for potential flicker issues before they manifest in the final encoded output.
Solution Approach 2:
The patent introduces an intermediary element - the provisional prediction image - that mediates between the reference frame and the current frame being encoded. This intermediary serves as a bridge that incorporates information from both past (reference frame) and future (succeeding frame) temporal contexts, smoothing out temporal variations that would otherwise cause flicker while maintaining encoding independence.
2Productivity
If inter frame prediction is used for encoding, then compression efficiency is improved, but temporal correlation and flicker issues arise
Solution Approach 1:
The patent merges forward prediction and backward prediction in bidirectional prediction to create a provisional prediction image. By combining these two prediction directions, the system maintains the temporal correlation benefits of inter frame prediction while compensating for flicker through the complementary nature of forward and backward predictions from different temporal references.
Solution Approach 2:
The patent changes the prediction parameter by using a weighted combination of forward and backward predictions rather than relying on a single prediction direction. This parameter change allows the system to optimize the balance between compression efficiency and temporal stability by adjusting the weights assigned to different prediction sources.
3Stability of the object's composition
If bidirectional prediction is used to reduce flicker, then image continuity is improved, but encoding complexity increases
Solution Approach 1:
The patent applies partial action by selectively using bidirectional prediction only for generating the provisional prediction image, while the final encoding can still use simpler unidirectional prediction methods. This partial application of the complex bidirectional technique maintains image continuity where needed without unnecessarily increasing encoding complexity across the entire encoding process.
Data Source
AI summary
The moving image encoding device maintains high encoding efficiency and performs encoding to reduce flicker occurrence in reproducing a moving image in moving image encoding processing by including an acquisition unit acquiring a succeeding frame of the target frame in encoding frame order when an encoding target frame is the inter picture in encoding a moving image including frames for which an intra or inter picture is determined; a determination unit determining whether an occurrence degree of flicker presumed in decoding the encoded moving image satisfies a criterion, based on characteristics of the target frame and the succeeding frame; and a generation unit 53 generating, when the determination result indicates that the criterion is satisfied, as a prediction target image used in encoding the target frame, a provisional prediction image based on the succeeding frame and a reference frame referred to by the target frame.