High-Resolution Video Encoding via Partition Segmentation and Additive Blending
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional video encoding methods for high-resolution videos suffer from low CPU utilization and real-time performance issues due to strong frame-level dependencies and reference correlations, leading to boundary effects and reduced video quality when scaling resolutions.
Innovation Solution
The method involves downsampling high-resolution images or videos, encoding the downsampled images, calculating differences, and encoding these differences to maintain image quality, allowing for efficient reconstruction without obvious boundary effects.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Manufacturing precision
If high-resolution videos are encoded using conventional methods, then video quality is maintained, but CPU utilization is low and real-time performance is poor due to strong frame-level dependencies
Solution Approach 1:
The patent divides the high-resolution video encoding task into multiple partitions (e.g., 4 partitions for 4K video). Each partition is encoded independently and simultaneously using multiple threads, eliminating frame-level dependencies and enabling true parallel processing. This segmentation allows the system to achieve high encoding speeds while maintaining video quality through independent partition encoding.
2Productivity
If high-resolution videos are divided into multiple low-resolution videos for parallel encoding, then encoding speed improves, but obvious boundary effects appear at partition boundaries
Solution Approach 1:
The patent merges multiple low-resolution partition encodings by upscaling each partition to the original high resolution and then combining them through additive blending. The formula used is: final_frame = upsampled_partition0 + upsampled_partition1 + upsampled_partition2 + upsampled_partition3. This merging process eliminates boundary effects by mathematically reconstructing the complete high-resolution image from its partitions, maintaining both encoding speed and video quality.
3Productivity
If multiple low-resolution videos are encoded separately, then parallel encoding capacity increases, but discontinuity effects at partition boundaries greatly reduce video quality
Solution Approach 1:
The patent introduces an intermediary upscaling and blending process between the separate low-resolution partition encodings and the final high-resolution output. Each partition is upscaled to high resolution, and then all upsampled partitions are added together to produce the final frame. This intermediary process acts as a mediator that reconciles the separately encoded partitions, eliminating discontinuity effects while preserving parallel encoding capacity.
4Productivity
If frame-level parallelism is used in high-resolution encoding, then encoding speed should improve, but strong reference dependency between adjacent frames limits parallel performance
Solution Approach 1:
The patent segments the video frame into multiple independent partitions that can be encoded simultaneously without reference to each other. By dividing the high-resolution frame into separate regions (e.g., four quadrants for 4K video), each partition becomes an independent encoding unit that eliminates inter-partition and inter-frame reference dependencies. This segmentation enables true frame-level parallelism where multiple threads can encode different partitions simultaneously without waiting for reference frames.
Data Source
AI summary
Image encoding and decoding methods and devices thereof are provided. The encoding method includes: performing downsampling on a first image to obtain a second image; encoding the second image to obtain a second image bit stream, and sending the second image bit stream to a decoding end; processing the second image to obtain a third image having a resolution the same as that of the first image; calculating a difference between the third image and the first image to obtain a first difference image; regulating pixel values of the first difference image to fall within a pre-set range, to obtain a second difference image; and encoding the second difference image to obtain a second difference image bit stream, and sending the second difference image bit stream to the decoding end to enable the decoding end to reconstruct the first image.