Pixel-Adaptive Image Restoration Filter for Noise Reduction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional image restoration filters, such as the Wiener filter, often result in over-restoration of pixels with smaller values and under-restoration of pixels with larger values, leading to increased noise and an inability to obtain high-quality images.
Innovation Solution
An image processing apparatus that includes an image restoration processing unit, a difference information generating unit, an adjustment coefficient setting unit, and a combining unit to calculate and apply correction difference information based on adjustment coefficients, allowing for tailored restoration adjustments to each pixel, thereby generating a high-quality image.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of operation
If uniform processing of the entire image using a conventional Wiener filter is performed, then the restoration process can be simplified, but over-restoration of pixels with smaller values occurs leading to increased noise and under-restoration of pixels with larger values
Solution Approach 1:
The patent applies local quality by setting different adjustment coefficients for different pixels based on their individual characteristics. Specifically, pixels with smaller values (darker regions) receive smaller adjustment coefficients to prevent over-restoration and noise amplification, while pixels with larger values (brighter regions) receive larger adjustment coefficients to achieve adequate restoration. This pixel-by-pixel customization resolves the contradiction by maintaining simplicity through automated coefficient selection while achieving high image quality through localized adaptation.
2Manufacturing precision
If adjustment coefficients are set for each pixel based on pixel values, then image quality is improved, but processing complexity increases
Solution Approach 1:
The patent employs parameter changes by dynamically adjusting the restoration filter parameters (adjustment coefficients) based on pixel values. The system changes the restoration strength parameter according to the intensity value of each pixel, using smaller coefficients for dark pixels and larger coefficients for bright pixels. This approach improves image quality through adaptive parameter selection while managing processing complexity by using a straightforward coefficient selection strategy based on pixel intensity thresholds.
3Manufacturing precision
If maximal restoration is applied to all pixels, then image sharpness is improved, but noise is significantly increased in regions with small pixel values
Solution Approach 1:
The patent applies local quality by differentiating the restoration strength applied to different regions of the image based on pixel values. Dark regions (small pixel values) receive minimal restoration with small adjustment coefficients to preserve noise-free characteristics, while bright regions (large pixel values) receive strong restoration with large adjustment coefficients to enhance sharpness. This localized approach resolves the contradiction by applying maximal restoration only where it benefits image quality without amplifying noise in sensitive regions.
Data Source
AI summary
To provide an image processing apparatus capable of obtaining a high-quality image, an image processing apparatus of the present invention includes an image restoration processing unit configured to perform a restoration process on an input image and generate a restored image, a difference information generating unit configured to calculate difference information between the restored image and the input image, an adjustment coefficient setting unit configured to be capable of setting a plurality of different adjustment coefficients for the input image, a correction difference information generating unit configured to generate correction difference information on the basis of the adjustment coefficients and the difference information, and a combining unit to combine the correction difference information with the input image and generate a restoration adjusted image.