Image Retargeting Quality Assessment via Frequency Domain Analysis
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current image retargeting algorithms often result in loss of image data and shape distortion, making it difficult to determine which algorithm produces the most appealing retargeted image for various screen sizes and resolutions.
Innovation Solution
A method for image retargeting quality assessment that compares original and retargeted images in both frequency and spatial domains, using frequency domain quality scores, shape distortion scores, and content quality scores to automatically select the highest quality retargeted image, with adaptive learning of quality score fusion models for improved accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If image retargeting algorithms are applied to resize images for various screen sizes, then images can be displayed on different devices, but image data loss and shape distortion occur
Solution Approach 1:
The patent replaces traditional spatial-domain image comparison methods with frequency-domain analysis using Fourier transforms. This substitution enables more accurate detection of image quality degradation by analyzing frequency components, thereby identifying data loss and distortion that spatial methods miss.
Solution Approach 2:
The patent changes the domain parameter from spatial to frequency domain for image analysis. By transforming images into the frequency domain and comparing spectral characteristics, the system can detect subtle data loss and distortion that are imperceptible in the spatial domain, thus resolving the contradiction between adaptability and information preservation.
2Manufacturing precision
If multiple image retargeting algorithms are developed to minimize data loss, then image quality may be improved, but no single algorithm is superior in all cases
Solution Approach 1:
The patent implements a feedback mechanism where the frequency-domain quality assessment model evaluates multiple retargeted images and provides quantitative quality scores. This feedback enables automatic selection of the best algorithm output without manual intervention, resolving the complexity of choosing among multiple algorithms while maintaining high image quality.
Solution Approach 2:
The system performs self-assessment of retargeted image quality through automated frequency-domain analysis. The quality assessment model independently evaluates each retargeted image and selects the best result, eliminating the need for external human judgment and simplifying the overall process despite multiple algorithms being involved.
3Device complexity
If traditional spatial domain comparison is used to assess retargeted images, then the process is simple, but it cannot accurately detect data loss and distortion
Solution Approach 1:
The patent substitutes traditional spatial-domain difference metrics with frequency-domain spectral analysis. By comparing Fourier transforms of original and retargeted images, the system achieves precise detection of data loss and distortion that spatial methods cannot detect, despite the increased computational complexity.
Solution Approach 2:
The patent transitions from two-dimensional spatial domain comparison to three-dimensional frequency domain analysis by introducing the frequency dimension. This dimensional change enables detection of subtle image degradation patterns that are invisible in spatial domain, significantly improving measurement precision while managing complexity through efficient signal processing.
Data Source
AI summary
A method of performing an image retargeting quality assessment comprising comparing an original image and a retargeted image in a frequency domain, wherein the retargeted image is obtained by performing a retargeting algorithm on the original image. The disclosure also includes an apparatus comprising a processor configured to perform an image retargeting quality assessment, and compare an original image and a retargeted image in a spatial domain, wherein the retargeted image is obtained by performing a retargeting algorithm on the original image, and wherein comparing the original image and the retargeted image in the spatial domain comprises comparing the original image and the retargeted image to determine an amount of shape distortion between the images.