AI In-Loop Video Filtering with Adaptive Pre-Processor Ratios
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional AI-based In Loop Filters (AILFs) struggle to achieve higher compression ratios and superior video quality due to their fixed ratios, limiting the efficiency of video compression processes.
Innovation Solution
Implementing a method and electronic device that utilize multiple pre-processors with varying channel ratios, selecting the optimal pre-processor based on extracted features for each frame, and embedding these ratios in headers or Coding Tree Units (CTUs) to enhance video encoding and decoding with AI-based In Loop Filtering (AILF).
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If a fixed ratio is used in AI-based In Loop Filter, then the filter structure is simple, but the compression ratio and video quality cannot be optimized
Solution Approach 1:
The patent implements dynamic ratio adjustment where the filter kernel size is adaptively modified based on extracted video features. The system transitions from a fixed ratio structure to a dynamic one that adjusts the ratio between different filter components (e.g., spatial vs. temporal filtering weights) according to the content characteristics, enabling optimization of compression ratio and video quality without excessive structural complexity
Solution Approach 2:
The patent changes the ratio parameter of the AI-based In Loop Filter dynamically based on extracted features from video frames. By modifying the filter ratio parameter according to content type, motion intensity, and other extracted characteristics, the system achieves adaptive optimization of compression efficiency and quality while maintaining a relatively simple base filter structure
2Productivity
If multiple pre-processors with varying channel ratios are implemented, then video compression efficiency is improved, but the device complexity increases
Solution Approach 1:
The patent segments the pre-processing function into multiple independent pre-processors, each with specific channel ratios tailored for different video characteristics. The system then selects and manages only the relevant pre-processors based on extracted features, which improves compression efficiency for specific content types while avoiding the complexity of managing all possible pre-processors simultaneously
Solution Approach 2:
The patent applies local quality by assigning different channel ratios to different pre-processors based on the specific needs of different video regions or content types. Instead of using a uniform pre-processor configuration, the system adapts the pre-processor characteristics locally to match the specific compression needs of different video segments, improving overall efficiency without requiring every possible configuration to be active
3Manufacturing precision
If feature extraction and pre-processor selection are performed for each frame, then compression quality is optimized, but processing time increases
Solution Approach 1:
The patent performs preliminary feature extraction and pre-processor selection based on frame characteristics before the main compression process. By pre-determining which pre-processor to use based on extracted features (e.g., motion vectors, texture analysis), the system avoids time-consuming trial-and-error during compression, thereby optimizing quality without significant time penalty
Solution Approach 2:
The patent implements feedback mechanisms where the extracted features from video frames are used to guide pre-processor selection and ratio adjustment. This feedback loop allows the system to adaptively choose the most efficient pre-processor based on actual content characteristics, improving compression quality while keeping processing time manageable through intelligent decision-making rather than exhaustive search
Data Source
AI summary
Embodiments herein provide a method and an electronic device for compressing a video for AI-based in loop filter (AILF). The method includes obtaining a video comprising a plurality of frames, wherein each frame of the pluralities of frames comprises a plurality of channels. Further, the method includes extracting at least one feature from each frame of the plurality of frames of the video. Further, the method includes selecting, at least one pre-processor from the multi pre-processors for the AILF based on the at least one feature from each frame of the plurality of frames. Further, the method includes generating an encoded video by encoding the image information from each channel of the plurality of channels of each frame of the plurality of frames using the at least one selected pre-processor.