Image Processing Apparatus Reducing Fixed-Pattern Noise
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing image processing methods fail to effectively reduce fixed-pattern noise in image sensor data, particularly in out-of-focus image groups and moving image data, leading to deteriorated image quality, especially when image magnification and reduction are small, and are unable to accurately handle multiplicative noise.
Innovation Solution
An image processing method that uses iterative calculations to determine optimal brightness changes for each pixel across multiple images, ensuring common changes for corresponding pixels, thereby reducing fixed-pattern noise and improving overall image quality.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If filter type methods are applied to an out-of-focus image group, then arbitrary viewpoint images or arbitrary out-of-focus images can be generated, but image quality deteriorates due to fixed-pattern noise
Solution Approach 1:
The patent applies preliminary action by performing fixed-pattern noise reduction processing on the out-of-focus image group before generating arbitrary viewpoint images or arbitrary out-of-focus images. This preprocessing step removes fixed-pattern noise from the input images, ensuring that the subsequent filter type methods operate on cleaned data, thereby preventing noise-induced quality deterioration while maintaining the ability to generate diverse viewpoint images.
2Stability of the object's composition
If image magnification and reduction are significantly small in an out-of-focus image group, then photography can be performed with a stable optical system, but fixed-pattern noise appears at approximately the same positions and becomes conspicuous
Solution Approach 1:
The patent converts the harmful effect of fixed-pattern noise into a beneficial outcome by exploiting the fact that the noise appears at consistent positions across images with small magnification changes. This positional consistency allows the algorithm to effectively identify and remove the fixed-pattern noise through comparative analysis, transforming the stability that causes noise visibility into an advantage for noise detection and elimination.
3Productivity
If a simplified process is used to reduce fixed-pattern noise in moving images, then processing speed is improved, but estimation accuracy of fixed-pattern noise is low
Solution Approach 1:
The patent applies dynamics by adapting the fixed-pattern noise reduction approach based on the specific characteristics of the input data. For moving images, the system dynamically selects and adjusts processing parameters to balance speed and accuracy, while for out-of-focus image groups, it employs more rigorous processing. This dynamic adaptation allows the system to optimize processing speed without sacrificing necessary accuracy in different应用场景.
Data Source
AI summary
An image processing method, includes: an input step in which a computer acquires data of a plurality of input images acquired by imaging performed using a same image sensor; and an optimization process step in which a computer determines an optimal solution of brightness change with respect to each input image by performing iterative calculations using an iterative method to improve overall image quality of the plurality of input images. In the optimization process step, an optimal solution of brightness change for each pixel of each input image is determined under a condition that common brightness change is performed with respect to pixels at a same position in respective input images.