Adaptive Temporal Filtering for Video Reference Frame Selection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing video encoding techniques do not always select the optimal reference frame for prediction, leading to suboptimal signal-to-noise ratios, especially when using alternative reference frames for backwards prediction.
Innovation Solution
The method generates candidate reference frames through adaptive temporal filtering based on relationships between frames, allowing for the selection of an optimal reference frame for encoding and decoding blocks, thereby increasing the likelihood of using the best reference frame for prediction.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If traditional reference frame selection is used, then encoding simplicity is maintained, but signal-to-noise ratio deteriorates
Solution Approach 1:
The patent segments the reference frame selection process into multiple stages: generating multiple candidate reference frames from different temporal positions, filtering candidates based on motion characteristics, and selecting the optimal reference frame. This segmentation transforms a single complex selection into manageable steps, improving signal-to-noise ratio while controlling encoding complexity.
Solution Approach 2:
The patent implements dynamic reference frame selection by adapting the filtering and selection process based on motion characteristics of the current block. The system dynamically adjusts which reference frames are candidates and applies motion-based filtering, allowing the encoding complexity to vary with scene complexity rather than being fixed, thus improving SNR without uniformly increasing complexity.
2Reliability
If multiple candidate reference frames are generated through temporal filtering, then signal-to-noise ratio is improved, but computational resources increase
Solution Approach 1:
The patent applies partial action by generating multiple candidate reference frames only when beneficial, using motion-based filtering to selectively process candidates, and abandoning excessive computation when motion characteristics indicate simple prediction suffices. This approach improves SNR through selective multi-candidate evaluation while avoiding unnecessary computational energy expenditure in scenes where simple reference frame selection is adequate.
3Measurement precision
If adaptive temporal filtering is applied to generate candidate reference frames, then prediction accuracy is improved, but encoding time increases
Solution Approach 1:
The patent performs preliminary action by pre-filtering candidate reference frames based on motion characteristics before the actual prediction process. By提前 (in advance) identifying and filtering suitable candidates using motion vectors and motion estimation, the system reduces the number of frames requiring detailed prediction processing, thereby improving prediction accuracy while limiting the time penalty to only the filtering stage rather than the entire encoding process.
Data Source
AI summary
Encoding or decoding blocks of video frames using multiple reference frames with adaptive temporal filtering can include generating one or more candidate reference frames by applying temporal filtering to one or more frames of a video sequence according to relationships between respective ones of the one or more frames and a current frame of the video sequence. A reference frame to use for predicting the current frame can be selected from the one or more candidate reference frames, and a prediction block can be generated using the selected reference frame. During an encoding operation, the prediction block can be used to encode a block of a current frame of the video sequence. During a decoding operation, the prediction block can be used to decode a block of a current frame of the video sequence.