Multi-scale Adaptive Image Fusion with Contrast Normalization
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current image fusion techniques, such as those employing Laplacian pyramid decomposition, struggle with high dynamic range and noise in images, particularly in low contrast areas, leading to reduced contrast and detail loss.
Innovation Solution
A computer-implemented method that generates Laplacian pyramid images, applies contrast normalization, and performs pixel-level fusion using a local salience measure to reduce aliasing artifacts, while noise coring and pre-processing techniques like histogram stretching and gamma correction enhance image quality.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If Laplacian pyramid fusion is used to select important image detail, then image detail selection is improved, but contrast is reduced and noise is amplified
Solution Approach 1:
The patent applies contrast normalization to the input images before performing Laplacian pyramid decomposition. This preliminary action ensures that the images have optimized contrast levels before the fusion process begins, preventing contrast loss during subsequent processing steps. The contrast normalization adjusts the dynamic range of each input image to match a target contrast level, ensuring that important details are preserved throughout the fusion process.
Solution Approach 2:
The patent dynamically adjusts the contrast parameter of input images by applying contrast normalization. This involves calculating a contrast gain factor for each pixel based on the local variance of pixel values, and then applying this gain to adjust the contrast. This parameter change ensures that images with varying contrast levels are normalized before fusion, preventing contrast reduction in the final fused image.
2Manufacturing precision
If Laplacian pyramid fusion is used to enhance video, then multi-frequency sharpening is improved, but noise is amplified
Solution Approach 1:
The patent converts the harmful effect of noise into a beneficial filtering process. By calculating local variance and using it to determine contrast gain factors, the method naturally suppresses noise in low-contrast regions while enhancing details in high-contrast regions. The contrast normalization process effectively filters out noise by adjusting gains based on local statistical properties, transforming what would be amplification of noise into selective noise suppression.
Solution Approach 2:
The patent applies different processing strengths to different regions of the image based on local characteristics. The contrast gain factor is calculated independently for each pixel based on local variance, allowing the method to apply stronger enhancement to regions with important details while applying weaker enhancement or suppression to regions dominated by noise. This local quality approach ensures that noise is not uniformly amplified across the entire image.
3Illumination intensity
If histogram stretching is applied to utilize input image dynamic range, then dynamic range utilization is improved, but localized low contrast areas are not adequately addressed
Solution Approach 1:
The patent transitions from global histogram stretching to local contrast normalization. Instead of applying a single contrast adjustment to the entire image, the method calculates contrast gain factors locally for each pixel based on the variance of pixel values in the neighborhood. This allows each local region to be optimized independently, addressing localized low contrast areas while maintaining overall dynamic range utilization.
Solution Approach 2:
The patent segments the image processing into local regions by calculating contrast metrics and applying gain factors on a pixel-by-pixel basis rather than globally. This segmentation allows the method to handle different contrast conditions in different parts of the image independently, effectively addressing localized low contrast areas that would be missed by global histogram stretching.
4Illumination intensity
If image fusion combines multiple video streams, then visibility enhancement is improved, but aliasing artifacts are introduced
Solution Approach 1:
The patent applies contrast normalization as a preliminary step before image fusion to ensure that all input images have optimized contrast levels. This preliminary action prevents contrast-related artifacts from being introduced during the fusion process and reduces the likelihood of aliasing artifacts by ensuring that important edge information is properly emphasized before combining multiple video streams.
Data Source
AI summary
A computer implemented method for fusing images taken by a plurality of cameras is disclosed, comprising the steps of: receiving a plurality of images of the same scene taken by the plurality of cameras; generating Laplacian pyramid images for each source image of the plurality of images; applying contrast normalization to the Laplacian pyramids images; performing pixel-level fusion on the Laplacian pyramid images based on a local salience measure that reduces aliasing artifacts to produce one salience-selected Laplacian pyramid image for each pyramid level; and combining the salience-selected Laplacian pyramid images into a fused image. Applying contrast normalization further comprises, for each Laplacian image at a given level: obtaining an energy image from the Laplacian image; determining a gain factor that is based on at least the energy image and a target contrast; and multiplying the Laplacian image by a gain factor to produce a normalized Laplacian image.