Compressed Spectral Imaging Noise Reduction via Low-Rank Approximation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Compressed spectral imaging produces encoded images with noise, such as thermal and shot noise, which are challenging to reduce effectively due to their distinct properties compared to general images, leading to potential distortion during reconstruction.
Innovation Solution
A noise reduction method involving a provisional image reconstruction, similar patch search, low-rank approximation, and patch integration to reduce noise in encoded images, utilizing a low-rank matrix reconstruction method based on weighted nuclear norm minimization.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of manufacture
If a general image noise reduction method is applied to encoded images from compressed spectral imaging, then the process is simple to implement, but the noise reduction effectiveness is poor due to significant property differences between general images and encoded images
Solution Approach 1:
The patent changes the fundamental parameters of the noise reduction approach by transitioning from general image processing to encoded image-specific processing. It introduces encoded image characteristics (such as multiplexed structures and wavelength-dependent PSF) as new parameters that must be considered, thereby adapting the noise reduction method to the specific properties of compressed spectral imaging data.
Solution Approach 2:
The patent segments the noise reduction process into multiple specialized stages: provisional image reconstruction, similar patch search, group generation, low-rank approximation, and patch integration. This segmentation allows each stage to address specific characteristics of encoded images, improving overall effectiveness while managing complexity through modular processing.
2Speed
If compressed spectral imaging is used to achieve high frame rate spectral images, then the frame rate is improved, but the amount of noise (thermal noise and shot noise) increases
Solution Approach 1:
The patent converts the harmful noise present in compressed spectral images into beneficial information by using the noise characteristics to guide the reconstruction process. The low-rank approximation leverages the statistical properties of the noise and signal to separate them, effectively using the noise pattern to improve the overall image quality rather than simply suppressing it.
Solution Approach 2:
The patent performs preliminary actions by first reconstructing a provisional image from the encoded image before conducting the similar patch search and noise reduction. This preliminary reconstruction creates a working model that guides subsequent processing steps, allowing noise reduction to be performed more effectively on the original encoded data.
3Loss of time
If the number of samples is reduced in compressed spectral imaging, then the acquisition time is reduced, but reconstruction quality deteriorates with artifacts and loss of fine features
Solution Approach 1:
The patent implements feedback mechanisms where the provisional image reconstruction is continuously refined based on the similarity of patches found during processing. The low-rank approximation uses feedback from the similar patch search to iteratively improve the reconstruction, adjusting the solution based on the consistency of identified similar regions until convergence is achieved.
Solution Approach 2:
The patent creates a composite processing approach that combines multiple techniques: compressed sensing reconstruction, similar patch matching, group generation, and low-rank approximation. This composite methodology integrates the strengths of each technique to achieve high reconstruction quality from reduced samples, overcoming the limitations of any single approach.
Data Source
AI summary
According to an aspect of the present invention, there is provided a noise reducing device including: a provisional image reconstructing unit that generates a provisional image by reconstructing an encoded image obtained through compressed spectral imaging; a similar patch search unit that acquires a plurality of similar provisional patches that are small regions including images similar to each other in the provisional image; a group generating unit that acquires a plurality of similar encoded patches that are small regions including images similar to each other in the encoded image based on information on positions where the similar provisional patches have been acquired; a low-rank approximation unit that performs low-rank approximation based on the plurality of similar encoded patches to acquire a patch in which noise on an image has been reduced in a region included in the similar encoded patches; and a patch integrating unit that generates an encoded image by integrating a plurality of patches in which noise has been reduced, depending on information on positions of the patches.