Image Stabilization via Cyclical Motion and Convolution Noise Correction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Image capture systems face challenges in correcting motion blur and spatial noise, with existing methods requiring complex calculations and assumptions that may lead to unsatisfactory results, and interrupting image capture for noise correction.
Innovation Solution
A method involving a stabilization module that applies a cyclical movement to the image capture device, allowing for continuous image capture while correcting spatial noise and motion blur by determining a global point spread function based on the defined movement, which is used to apply a convolution product to intermediate images, thereby reducing noise and improving image quality.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If deconvolution methods are used to correct motion blur, then motion blur correction is achieved, but complex calculations and prior assumptions are required which may lead to unsatisfactory results
Solution Approach 1:
The patent replaces complex mathematical deconvolution operations with a simpler convolution-based approach. Instead of performing computationally intensive deconvolution to remove motion blur, the system applies a convolution operation with a pre-determined point spread function that directly corrects the blur effect, significantly reducing calculation complexity while maintaining correction effectiveness
Solution Approach 2:
The system determines the point spread function representing motion blur in advance, before the actual image correction is needed. This preliminary determination of the blur characteristics allows the subsequent correction step to use a pre-characterized model, avoiding the need for complex real-time deconvolution calculations and prior assumptions about the blur type
2Object-affected harmful factors
If inertial stabilization is used to reduce motion blur, then motion blur is avoided, but fixed spatial noise appears in the captured images
Solution Approach 1:
The patent converts the harmful fixed spatial noise into a correctable artifact by recognizing its deterministic nature. The system captures multiple images during the cyclical movement, and the fixed spatial noise appears consistently across these images. By using convolution with an appropriately designed point spread function, the system can selectively enhance or suppress these consistent patterns, effectively converting the previously harmful noise into a feature that can be manipulated to improve overall image quality
Solution Approach 2:
The system employs periodic cyclical movement of the image capture device, capturing multiple images at different positions within the cycle. This periodic action creates a structured pattern in the captured images where fixed spatial noise appears at consistent locations across the sequence, enabling its identification and correction through convolution operations while the motion blur is minimized due to the controlled, repetitive nature of the movement
3Measurement precision
If spatial noise correction is performed by capturing a black body image, then spatial noise is corrected, but image capture process is interrupted
Solution Approach 1:
The patent makes the image capture system multi-functional by designing a correction method that works with normal scene images rather than requiring separate black body calibration images. The same convolution-based correction process that removes motion blur also addresses fixed spatial noise, allowing the system to perform both motion correction and spatial noise correction using the captured scene images themselves, eliminating the need to interrupt capture for separate calibration
Solution Approach 2:
The system maintains continuous image capture throughout the correction process. By using the cyclical movement of the capture device and applying convolution operations to the sequence of captured images, the correction is performed continuously on the actual scene data without requiring pauses for separate calibration shots. This ensures uninterrupted image capture while still achieving both motion blur reduction and spatial noise correction
Applied Scientific Principles
This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.
Function Achieved in This Case
Enables effective correction of spatial noise and motion blur without interrupting image capture, improving image quality by reducing temporal and spatial noise through controlled movement of the image capture device.
Implementation Method 1
said viewing axis being animated by a defined, cyclic movement
Implementation Method 2
obtain a corrected captured image of the scene by applying a convolution product between said intermediate image and a point spreading function
Data Source
Figure 1~4
Figure 2
AI summary
In a system for capturing a series of images of a scene comprising an image capture device having a matrix (52) of flux detectors which is oriented along a sighting axis (z), comprising a stabilizing module which stabilizes the sighting axis of the captured images, the sighting axis being instilled with a defined, cyclic, motion, the following steps are applied to a series of successive images (N1-K) captured during a cycle. Steps /i/ and /ii/ are repeated from the first image captured up to the last image captured of the series and an intermediate image (Iintermediate) is obtained - /i/ registering (21) a subsequent image (Ni+i) on a preceding image (Ni) as a function of the defined motion; - /ii/ correcting (22) spatial noise relating to the flux detectors and affecting the common part of the images. Next, a corrected captured image of the scene is obtained (25) by applying a convolution product (26) between the intermediate image and a correction point spreading function (PSFglobal) which is determined on the basis of the defined motion.