Adaptive Frequency Lifting for Image Detail Enhancement
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Ultra-high definition (UD) displays struggle to maintain image quality when showing lower resolution content, as regular image upscaling or interpolation fails to provide sufficient sharpness and fine details, leading to degraded image quality.
Innovation Solution
A method that determines enhancement information based on frequency characteristics and texture information of input images, using a processor to transform and shift coefficients in a matrix, perform frequency lifting, and mix the enhancement information with the input image to generate an enhanced image with improved details.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Manufacturing precision
If regular image upscaling or interpolation is used to display lower resolution content on UD devices, then the content can be displayed, but the image sharpness and fine details are insufficient, leading to degraded image quality
Solution Approach 1:
The patent transforms the image from spatial domain to frequency domain using 2D DFT, then modifies frequency parameters by lifting selected frequency components to higher frequencies. This parameter transformation and modification enables creation of high-frequency details that were missing in the original low-resolution image, thereby improving image sharpness and quality without requiring higher resolution input
Solution Approach 2:
The patent transitions from spatial domain processing to frequency domain processing, adding a frequency dimension to the image representation. By operating in the frequency domain and then transforming back to spatial domain, the system creates new high-frequency information that enhances image details and sharpness, resolving the quality degradation issue
2Manufacturing precision
If frequency lifting is performed to create high-frequency components for image enhancement, then image sharpness and details are improved, but the computational complexity increases
Solution Approach 1:
The patent applies frequency lifting selectively to specific frequency components rather than transforming the entire image uniformly. By identifying and lifting only the necessary frequency components that contribute to image details, the system achieves effective image enhancement while reducing unnecessary computational overhead compared to full-image frequency transformations
Solution Approach 2:
The patent divides the frequency spectrum into different components and selectively processes specific frequency regions. By segmenting the frequency domain and applying lifting operations only to relevant components, the system optimizes computational efficiency while maintaining image quality improvement
Data Source
Figure 1
Figure 2
Figure 3~4
AI summary
Image details are created for an image. An image is processed to obtain image information. Detected texture samples are processed to obtain texture information. A processor device is used for transforming received image information for obtaining frequency domain information. The texture information is used for determining a lifting factor. A frequency distribution is determined for the frequency domain information using the lifting factor for creating particular frequency. An inverse transformation is performed on an updated frequency distribution for creating output image blocks. The output image blocks are combined to create image details for a lifted image. The lifted image is mixed with the image.