Spatiotemporal Volume Processing for Atmospheric Turbulence Suppression
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional noise filtering techniques fail to effectively suppress both random and fixed pattern noise in images, especially in low-cost sensors with narrow pixel-pitch or low signal-to-noise ratios, and struggle with distortion caused by atmospheric turbulence, leading to artifacts and loss of high-frequency content.
Innovation Solution
The method involves constructing spatiotemporal volumes from image blocks along motion trajectories, smoothing these trajectories to suppress random displacements, and applying decorrelating transforms to suppress blurring effects, allowing for accurate modeling and filtering of both random noise and fixed pattern noise components, as well as distortion caused by atmospheric turbulence.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional noise filtering techniques (averaging or smoothing operations) are applied to suppress noise, then random noise may be reduced, but fixed pattern noise becomes more visible and structured artifacts are produced
Solution Approach 1:
The patent segments the image into multiple overlapping patches and processes each patch independently through spatiotemporal volumes. This segmentation allows different filtering strategies to be applied to different regions, preventing the propagation of structured artifacts while maintaining noise suppression effectiveness.
Solution Approach 2:
The patent transitions from 2D spatial filtering to 3D spatiotemporal filtering by incorporating the time dimension. Spatiotemporal volumes are constructed by stacking multiple frames along the temporal dimension, enabling noise suppression that exploits temporal redundancy without producing spatial artifacts.
2Measurement precision
If conventional techniques (bispectrum imaging, lucky imaging, temporal averaging) are applied to address atmospheric turbulence distortion, then some distortion may be reduced, but high temporal frequency content is lost and motion blur/ghosting effects occur in scenes with motion
Solution Approach 1:
The patent employs dynamic patch tracking that adapts to scene motion by computing motion vectors for each patch across frames. This dynamic approach allows the spatiotemporal volumes to follow moving objects, preserving high temporal frequency content while suppressing turbulence-induced distortion through collaborative filtering in the spatiotemporal domain.
Solution Approach 2:
The patent changes the filtering parameters adaptively based on local motion characteristics. By adjusting the spatiotemporal volume construction and filtering strength according to motion vectors and turbulence estimates, the system maintains high-frequency content for moving objects while suppressing distortion in static regions.
3Manufacturing precision
If spatiotemporal volumes are constructed and decorrelating transforms are applied to suppress blurring effects, then image clarity is improved, but device complexity increases
Solution Approach 1:
The patent divides the image into overlapping patches and processes each patch through its own spatiotemporal volume independently. This segmentation reduces the overall computational complexity compared to processing the entire image at once, while still achieving improved image clarity through localized decorrelating transforms.
Solution Approach 2:
The patent applies decorrelating transforms selectively to spatiotemporal volumes rather than to all image data uniformly. By focusing computational resources on regions with significant turbulence distortion and using partial processing of spatiotemporal volumes, the system achieves improved clarity without excessive processing complexity.
Data Source
AI summary
Various techniques are disclosed to suppress distortion in images (e.g., video or still images), such as distortion caused by atmospheric turbulence. For example, similar image blocks from a sequence of images may be identified and tracked along motion trajectories to construct spatiotemporal volumes. The motion trajectories are smoothed to estimate the true positions of the image blocks without random displacements/shifts due to the distortion, and the smoothed trajectories are used to aggregate the image blocks in their new estimated positions to reconstruct the sequence of images with the random displacements/shifts suppressed. Blurring that may remain within each image block of the spatiotemporal volumes may be suppressed by modifying the spatiotemporal volumes in a collaborative fashion. For example, a decorrelating transform may be applied to the spatiotemporal volumes to suppress the blurring in a transform domain, such as by alpha-rooting or other suitable operations on the coefficients of the spectral volumes.