Accented Image Data Generation via Spectral-Edge Decomposition
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current imaging technologies face challenges in visualizing high-dimensional image data, such as hyperspectral and multispectral images, which capture more than three channels, as human observers can only perceive three color dimensions, leading to loss of information and inaccurate color representation when trying to display these images.
Innovation Solution
A method that decomposes images into spectral and edge components using singular value decomposition, combining edge components from one image with spectral components from another to create an accented image that retains useful edge information and compresses data into a lower-dimensional RGB format, allowing for accurate feature identification.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of operation
If high-dimensional image data (hyperspectral/multispectral) is displayed using only visible spectrum channels, then the image can be viewed by human observers, but information from additional spectral modalities (infrared, ultraviolet) is lost
Solution Approach 1:
The patent maps spectral information from high-dimensional space (multiple spectral bands including infrared and ultraviolet) into the three-dimensional RGB color space by decomposing the spectral data and reconstructing it through a transformation process that preserves key spectral characteristics while making the data compatible with human visual perception
Solution Approach 2:
The patent extracts key spectral components from the high-dimensional image data by decomposing it into basis spectra and their corresponding coefficients, then uses only the essential spectral information needed for accurate color representation while discarding redundant data
2Loss of information
If false-colour images are created by blending all channels together, then information from all modalities is preserved, but the colours assigned to objects are markedly different from true colours
Solution Approach 1:
The patent changes the spectral parameters by transforming the high-dimensional spectral data into a reduced set of basis spectra that capture the essential spectral variations, then uses these transformed parameters to compute accurate RGB color values that reflect true object colors rather than arbitrary false-color assignments
3Productivity
If hyperspectral images are compressed to RGB format, then the image can be visualized and stored efficiently, but the large amount of spectral data is lost
Solution Approach 1:
The patent segments the high-dimensional spectral data into multiple basis spectra and their corresponding coefficient maps, allowing the data to be stored and processed in a compressed format while preserving the essential spectral information needed for accurate color reconstruction and analysis
Data Source
Figure 1
Figure 2
Figure 3a~3c
AI summary
A method and system for producing accented image data for an accented image is disclosed. The method includes decomposing each of a first and a second image into a gradient representation which comprises spectral and edge components. The first image comprises more spectral dimensions than the second image. The edge component from the first image is combined with the spectral component from the second image to form a combined gradient representation. Accented image data for the accented image is then generated from data including the combined gradient representation.