Shakiness Correction Using Parallel Filter Modes
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional shakiness correction techniques in digital imaging apparatuses over-correct when intentionally moving a camera, leading to reduced accuracy and introduce delays due to the need to switch between hand shaking and panning operations.
Innovation Solution
A method and apparatus that generate and selectively apply two types of filtering data, one for panning mode and another for capture mode, allowing for parallel processing and immediate mode switching without changing filter factors, enabling precise and rapid shakiness correction.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If a single filter is used for both hand shaking and panning operations, then the device complexity is reduced, but the manufacturing precision deteriorates due to over-correction
Solution Approach 1:
The patent divides the filter system into two separate filters: a first filter for hand shaking correction and a second filter for panning operation correction. Each filter has its own filter factor optimized for its specific function, eliminating the over-correction problem that occurs when a single filter is used for both purposes.
Solution Approach 2:
The patent applies different filter factors to different operational modes. The first filter factor is specifically optimized for hand shaking correction while the second filter factor is optimized for panning operations, ensuring that each operation receives the appropriate level of correction without interfering with the other.
2Manufacturing precision
If filter factors are switched between hand shaking and panning modes, then the manufacturing precision is improved, but the loss of time increases due to filter output stabilization delay
Solution Approach 1:
The patent maintains both filters operating simultaneously in parallel, with each filter continuously processing data for its respective mode. This preliminary preparation of both filter outputs eliminates the delay that would otherwise occur when switching between modes, as both correction types are already computed and ready for immediate application.
Solution Approach 2:
The patent ensures continuous operation of both filters without interruption or switching delays. By maintaining both hand shaking correction and panning correction filters active simultaneously, the system provides uninterrupted correction capability for whichever mode is currently active, eliminating the stabilization delay inherent in sequential filter switching.
3Adaptability or versatility
If correction is performed for both hand shaking and panning using the same filter, then the adaptability is improved, but the manufacturing precision deteriorates due to inability to distinguish intentional movement from hand shaking
Solution Approach 1:
The patent segments the correction system into distinct filters for different movement types. The first filter specifically targets hand shaking characteristics while the second filter handles intentional panning movements, allowing the system to adapt to different modes while maintaining high precision by treating each movement type separately.
Solution Approach 2:
The patent applies specialized correction characteristics to each filter type. The hand shaking filter is optimized to detect and correct high-frequency involuntary movements, while the panning filter is optimized for lower-frequency intentional camera movements, enabling accurate distinction and correction of different movement types simultaneously.
Data Source
AI summary
A shakiness correcting method includes: generating, based on motion sensing data, first filtering data to be applied to a first mode; generating, based on the motion sensing data, second filtering data to be applied to a second mode; and selectively performing shakiness correction on one of the first filtering data and the second filtering data according to mode selection information, wherein the generating of the first filtering data and the generating of the second filtering data are performed in a mutually parallel manner.