Content Adaptive Image Restoration for High Definition Displays
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing image processing systems for high definition displays face issues with noise reduction and texture preservation when converting low resolution video to high resolution, leading to irreversible detail loss, noise remnants, and jagged edges due to inadequate adaptation to the nature of the input video.
Innovation Solution
A content adaptive resolution enhancer (CARE) system that uses a texture estimator and noise discriminator to generate adaptive kernels, a 2D adaptive sharpener to filter luminance signals, and a scaler to vertically and horizontally scale images, ensuring adaptive processing across all stages to enhance image quality.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Object-affected harmful factors
If noise reduction is performed in the pre-scaler domain, then noise is reduced, but fine textures and details are lost irreversibly
Solution Approach 1:
The patent applies preliminary action by performing noise reduction in the pre-scaler domain before scaling operations. The pre-scaler domain image restoration block processes the low resolution video signal to remove noise before it is scaled up, preventing noise from being amplified in subsequent processing stages.
Solution Approach 2:
The patent segments the image processing into distinct domains: pre-scaler domain for noise reduction, scaling domain for resolution conversion, and post-scaler domain for edge enhancement. This segmentation allows each domain to perform its specific function optimally without interfering with others, preventing irreversible detail loss while maintaining noise reduction benefits.
2Device complexity
If a single wide band polyphase filter is used for scaling, then scaling is simplified, but noise remnants are scaled into the high resolution domain
Solution Approach 1:
The patent segments the scaling process into multiple stages with separate vertical and horizontal scaling blocks, each using polyphase filtering. This segmentation allows independent optimization of each scaling dimension and enables better control over noise propagation during the scaling process.
Solution Approach 2:
The patent changes the filter parameters dynamically by using polyphase filters with adjustable coefficients in the vertical and horizontal scaling blocks. This allows the system to adapt the scaling characteristics to minimize noise remnants while maintaining computational efficiency.
3Device complexity
If separable vertical and horizontal sharpness is applied, then processing is simplified, but jaggedness is amplified in slant edges
Solution Approach 1:
The patent segments the sharpness enhancement into separate vertical and horizontal components processed by distinct blocks. This segmentation allows independent optimization of vertical and horizontal edge enhancement while maintaining the ability to handle slant edges through coordinated processing in both domains.
Solution Approach 2:
The patent applies local quality by adapting the sharpness enhancement to different regions and orientations. The post-scaler edge domain image enhancement block selectively enhances edges based on their orientation and characteristics, preventing jaggedness in slant edges while maintaining simplicity in processing.
4Device complexity
If image enhancement is performed over the entire image, then processing is uniform, but stray noise appears in texture regions
Solution Approach 1:
The patent applies local quality by differentiating processing for different image regions. The post-scaler edge domain image enhancement block analyzes local characteristics and applies enhanced processing only where needed (at edges) while maintaining original processing for texture regions, preventing stray noise generation.
Solution Approach 2:
The patent changes processing parameters dynamically based on local image characteristics. The system adjusts enhancement strength and filter characteristics according to whether a region contains edges, textures, or noise, allowing uniform processing strategy to adapt to local requirements and avoid stray noise in texture regions.
Data Source
AI summary
An image processor includes generates a content adaptive kernel from an image block with noise of a luminance component signal with a low resolution. The content adaptive kernel is convolved with the luminance component signal. A noise signal and an extracted texture which excludes noise are generated. The luminance component signal is filtered as function of the noise signal to generate an enhanced luminance component signal. Horizontal and vertical scaling is performed on the enhanced luminance component signal, the extracted texture, and the luminance component signal, with the luminance component signal adaptively scaled as a function of the extracted texture. The horizontally and vertically scaled enhanced luminance component signal, extracted texture and luminance component signal are then combined to generate an output luminance component signal with a high resolution.