Image Scaling with Reference Region and Quality Metric Selection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing image rendering techniques often result in visual quality deterioration when displaying images with different resolutions and aspect ratios, particularly when scaling images to fit a display area, leading to distortion and inefficient use of the display space.
Innovation Solution
A method that defines a reference region in the source image, forms candidate image regions with a target aspect ratio, scales each region to form candidate target images, and selects the image with the best quality metric value to ensure optimal visual quality, excluding unfavorable scaling factors and using metrics like blur and structural similarity to evaluate image quality.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Stability of the object's composition
If the original image's aspect ratio is preserved during scaling, then the image maintains its original proportions, but black bands are displayed and part of the thumbnail is not used
Solution Approach 1:
The image transformation process is segmented into multiple candidate transformations, each preserving aspect ratio differently. The system evaluates multiple segmentation options of the source image to find the optimal region that balances aspect ratio preservation with display area utilization.
Solution Approach 2:
The solution moves from a single-dimensional aspect ratio preservation to a two-dimensional optimization by evaluating multiple candidate regions and transformations. It considers both horizontal and vertical scaling factors simultaneously to find the optimal transformation that utilizes display area while maintaining visual quality.
2Area of stationary object
If the image is transformed to fill the square thumbnail, then the entire surface is used to display the image, but the image has vertical stretch distortion
Solution Approach 1:
The system changes the parameters of transformation by evaluating multiple candidate transformations with different scaling factors. Instead of using a single fixed transformation, it varies the scaling parameters horizontally and vertically to find the combination that minimizes distortion while maximizing display area utilization.
Solution Approach 2:
The transformation approach is made dynamic by selecting from multiple candidate transformations based on quality metrics. The system adapts the transformation parameters dynamically according to the specific image content and display requirements, rather than applying a static transformation rule.
3Ease of manufacture
If sophisticated scaling algorithms are used, then the scaling quality is improved, but visual quality deterioration is still perceived
Solution Approach 1:
The system incorporates feedback through quality metrics that evaluate each candidate transformation. By measuring visual quality indicators such as blur and structural similarity, the system receives feedback on transformation quality and uses this to select the optimal transformation, thereby improving visual quality beyond what sophisticated scaling alone can achieve.
Solution Approach 2:
The system performs preliminary evaluation of multiple candidate transformations before final rendering. By pre-calculating quality metrics for each candidate and selecting the best one in advance, it avoids the visual quality deterioration that occurs with post-processing adjustments or simple scaling algorithms.
4Manufacturing precision
If multiple candidate regions are evaluated with quality metrics, then the optimal transformation is selected for best visual quality, but the processing time and complexity increase
Solution Approach 1:
The system applies partial evaluation by considering multiple candidate transformations but not exhaustively evaluating all possible transformations. It uses quality metrics to evaluate a sufficient number of candidates to ensure optimal visual quality without unnecessary excessive processing, balancing quality improvement with computational efficiency.
Data Source
AI summary
A source image is transformed into a destination image having a target aspect ratio. A reference region in the source image is defined. An extended region of interest of the source image having the target aspect ratio and containing the reference region is defined. A set of candidate image regions of increasing resolutions from the extended region of interest is determined, each having the target aspect ratio and containing the reference region. Candidate image regions are scaled to form a candidate target images. A quality metric is used to select a target image providing the best quality metric value.