Super-Resolved Satellite Image Fusion via Subpixel Alignment
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current satellite imaging technologies, particularly those using small and low-cost satellites like SkySat, suffer from low image resolution and high noise due to movement, resulting in a short acquisition time and low signal-to-noise ratio, which complicates the acquisition of clear and detailed images.
Innovation Solution
A method and system for creating super-resolved images from sequences of images taken by overhead optical imaging devices, involving geometric deformation estimation, subpixel position calculation, and image fusion using techniques like wavelet, Fourier, and spline interpolation to improve spatial resolution and signal-to-noise ratio.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If a longer exposition time is used to improve signal-to-noise ratio, then the signal-to-noise ratio increases, but image blur occurs due to satellite movement
Solution Approach 1:
The patent performs geometric deformation estimation and subpixel position calculation before image fusion to pre-align multiple images. This preliminary registration ensures that images captured at different times can be accurately combined without blur, allowing longer effective exposure through multiple frames while maintaining sharpness through predictive alignment
Solution Approach 2:
The patent combines multiple images captured at different times into a single super-resolved image through fusion algorithms. By merging multiple short-exposure images that are geometrically aligned, the system achieves both high signal-to-noise ratio (through accumulation of signal across frames) and sharpness (through precise geometric correction before fusion)
2Shape
If a short acquisition time is used to avoid image blur, then image sharpness is maintained, but signal-to-noise ratio decreases
Solution Approach 1:
The patent processes a continuous sequence of images captured during satellite passage, utilizing all available frames in the sequence. By continuously processing multiple frames through geometric correction and fusion, the system accumulates signal from each short-exposure frame while maintaining sharpness through ongoing geometric alignment, effectively extending the useful acquisition period without increasing individual exposure time
Solution Approach 2:
The patent performs preliminary geometric deformation estimation and subpixel position calculation for each frame before fusion. This pre-processing ensures that even though each individual frame has short exposure time for sharpness, the cumulative effect of multiple pre-aligned frames produces high signal-to-noise ratio in the final super-resolved image
3Measurement precision
If multiple images are processed to improve resolution, then spatial resolution and signal-to-noise ratio increase, but processing complexity increases
Solution Approach 1:
The patent segments the image processing into distinct modular steps: geometric deformation estimation, subpixel position calculation, and fusion algorithm application. This segmentation allows each module to be optimized independently and processed efficiently, reducing overall computational complexity while achieving super-resolution through systematic multi-stage processing
Solution Approach 2:
The patent moves from pixel-level processing to subpixel-level processing by calculating subpixel positions for each image. This dimensional refinement (from discrete pixel grid to continuous subpixel coordinates) enables more precise alignment and fusion, achieving higher spatial resolution without proportionally increasing processing complexity through efficient interpolation methods
Data Source
AI summary
Methods, devices and non-transitory computer-readable storage medium for processing a sequence of respective top view images of a same terrestrial location are provided. One of the method may comprise choosing one image, called reference image, among the respective top view images, estimating for each respective top view image a respective geometric deformation between the respective top view image and the reference image, computing by the respective geometric deformations respective subpixel positions of the respective top view images relative to one high-resolution coordinate system, interpolating at the respective subpixel positions to sample at least part of at least some of the respective top view images on a prescribed grid to obtain a high-resolution image.