Component-Based Image Preprocessing for Lower-Bitrate Video Quality
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional video encoding pipelines suffer from suboptimal visual quality degradation due to independently optimized preprocessing algorithms that are not designed to work cohesively, leading to increased artifacts and higher bitrates or lower quality in reconstructed videos.
Innovation Solution
Implementing trained machine learning models for component-based image preprocessing in a luma-chroma color space to reduce reconstruction errors, optimizing the aggregation of preprocessing operations within a video encoding pipeline.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of manufacture
If independently optimized preprocessing algorithms are used in conventional video encoding pipelines, then each algorithm can be designed and optimized individually, but the aggregated preprocessing operations produce increased visual quality degradation and artifacts
Solution Approach 1:
The patent combines multiple independently optimized preprocessing algorithms into a unified, cohesively designed preprocessing pipeline. By merging chroma subsampling, downscaling, and spatial denoising operations into an integrated system, the patent eliminates the cumulative visual quality degradation that occurs when these algorithms are applied sequentially and independently, thereby resolving the contradiction between ease of independent algorithm design and overall visual quality preservation
2Quantity of substance
If lossy encoding algorithms are applied to preprocessed video, then the bitrate is reduced and file size decreases, but visual quality degradation and artifacts increase
Solution Approach 1:
The patent applies a cohesively designed preprocessing pipeline before lossy encoding to preemptively optimize the video data structure and reduce information that would otherwise be lost during compression. By performing preliminary actions such as intelligent chroma subsampling, adaptive downscaling, and noise suppression in a unified framework, the patent minimizes the visual quality degradation that would otherwise occur during subsequent lossy encoding, thereby resolving the contradiction between bitrate reduction and visual quality preservation
3Adaptability or versatility
If multiple preprocessing algorithms are applied sequentially, then different types of image preprocessing are performed, but the cumulative effect increases the number and severity of artifacts
Solution Approach 1:
The patent implements a dynamic, adaptively controlled preprocessing pipeline where the strength and application of each preprocessing operation (chroma subsampling, downscaling, spatial denoising) are adjusted based on the characteristics of the input video and the presence of other preprocessing operations. This dynamic coordination prevents the cumulative artifact generation that occurs when fixed, independently optimized algorithms are applied sequentially, thereby resolving the contradiction between preprocessing versatility and artifact reduction
Data Source
AI summary
In various embodiments, an image preprocessing application preprocesses images. To preprocess an image, the image preprocessing application executes a trained machine learning model on first data corresponding to both the image and a first set of components of a luma-chroma color space to generate first preprocessed data. The image preprocessing application executes at least a different trained machine learning model or a non-machine learning algorithm on second data corresponding to both the image and a second set of components of the luma-chroma color space to generate second preprocessed data. Subsequently, the image preprocessing application aggregates at least the first preprocessed data and the second preprocessed data to generate a preprocessed image.