Saliency-Based Video Compression Tile Quality
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current video compression technologies fail to efficiently reduce video data size while maintaining quality, especially with the rise of high-resolution and immersive video formats like 360-degree and 8K VR, due to their focus on single codec implementations and neglect of perceptual cues like saliency.
Innovation Solution
Integration of saliency-based compression techniques using neural networks to predict saliency maps and adjust video tile quality settings, allowing for significant bitrate reduction and power savings in video storage and playback, compatible with existing codecs like HEVC and AV1.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Manufacturing precision
If traditional video compression codecs are used, then video quality is maintained, but video file size becomes excessively large and storage costs increase
Solution Approach 1:
The patent applies local quality by dividing the video frame into multiple tiles and assigning different quality levels to different tiles based on their saliency values. High-salience regions (where viewers are likely to focus attention) are encoded with higher quality, while low-salience regions are encoded with lower quality. This resolves the contradiction by maintaining perceived video quality in important areas while reducing overall file size through selective quality reduction in less important areas.
Solution Approach 2:
The patent changes the quality parameter dynamically based on saliency information. Instead of applying a uniform quality setting across the entire video, the system adjusts the quality parameter for each tile according to its saliency value. This allows the video compression to adapt to human visual perception characteristics, maintaining quality where it matters most while achieving significant file size reduction overall.
2Manufacturing precision
If high-resolution and immersive video formats (360-degree, 8K VR) are adopted, then video quality and immersion are improved, but storage requirements and power consumption increase significantly
Solution Approach 1:
For high-resolution immersive videos, the patent applies local quality by identifying and prioritizing salient regions within the large video frames. Instead of uniformly compressing the entire high-resolution content, the system maintains high quality only in regions where viewers are likely to focus attention, while applying more aggressive compression to less important regions. This significantly reduces storage requirements and power consumption while preserving the immersive experience in critical areas.
3Ease of manufacture
If uniform quality compression is applied to all video regions, then implementation is simple, but bitrate efficiency is suboptimal and file sizes are larger than necessary
Solution Approach 1:
The patent segments the video frame into multiple tiles and processes each tile independently with its own quality setting based on saliency. This segmentation approach maintains relative implementation simplicity by using standard video coding techniques for each tile, while dramatically improving bitrate efficiency through perceptually-driven differential compression across tiles.
Data Source
AI summary
Examples of systems and methods described herein may provide saliency-based video compression. A saliency map associated with a video may be generated and/or provided. A tile configuration may be selected for the video and quality settings assigned to each tile in accordance with the saliency map. The video may then be compressed (e.g., encoded) in tiles in accordance with the quality settings. Compressed videos may be stored together with saliency metadata, facilitating storage management and/or re-compression.