Multispectral Guide Image Noise Reduction for Low-Light Imaging
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Traditional spatiotemporal noise reduction techniques are less reliable in low-light conditions due to unreliable structural correspondences, leading to increased noise and motion artifacts, especially when capturing multispectral image data.
Innovation Solution
The use of multispectral imaging sensors that combine visible and non-visible light signals to create a 'multispectral guide image' for noise reduction, employing noise normalization and non-local means algorithms, and noise-guided image frame accumulation to improve noise reduction accuracy, particularly in low-light environments.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional spatiotemporal noise reduction techniques are used in low-light conditions, then processing speed is maintained, but noise reduction accuracy deteriorates due to unreliable structural correspondences
Solution Approach 1:
The patent combines multiple spectral bands (visible and non-visible light) to create a composite guide image. This merging of spectral information provides more reliable structural correspondences for noise reduction in low-light conditions, resolving the contradiction between maintaining processing speed and improving noise reduction accuracy.
Solution Approach 2:
The patent introduces a multispectral guide image as an intermediary element that facilitates accurate correspondence matching. This guide image, derived from combined spectral data, serves as a mediator that enables reliable structural comparisons even when individual spectral bands have poor signal-to-noise ratios.
2Object-affected harmful factors
If traditional noise reduction methods are applied to multispectral image data, then processing complexity is kept simple, but noise and motion artifacts increase in low-light environments
Solution Approach 1:
The patent segments the processing into distinct stages: creating a multispectral guide image, calculating blending weights based on the guide image, and applying these weights to denoise individual spectral bands. This segmentation allows complex multispectral processing to be broken down into manageable steps, reducing artifacts while controlling processing complexity.
Solution Approach 2:
The patent extends traditional 2D spatial noise reduction into the spectral dimension by incorporating multiple spectral bands. The multispectral guide image utilizes information across different spectral dimensions to guide denoising, effectively adding a new dimension to the noise reduction process that improves artifact reduction without excessive complexity.
Data Source
AI summary
Devices, methods, and non-transitory program storage devices for spatiotemporal image noise reduction are disclosed, comprising: maintaining an accumulated image in memory; and obtaining a first plurality of multispectral images (e.g., RGB-IR images). For each image in the first plurality of multispectral images, the method may: calculate a multispectral guide image for the current image; calculate blending weights for the current image; apply the calculated blending weights to each channel of the current image to generate a denoised current image; and update the accumulated image based on pixel differences between the denoised current image and the accumulated image. In some embodiments, additional images (e.g., the accumulated image and/or other images captured prior to or after a given current image) may also be included in the denoising operations for a given current image. Finally, the method may generate a denoised output image for each input image, based on the updated accumulated image.