Image Sensor Fixed Pattern Noise Reduction via Center-Weighted Filtering

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing image processing technologies fail to effectively reduce fixed patterns generated by repeating cycles in image sensors without compromising image detail or quality.

Innovation Solution

An image processing device that uses a filter with coefficients weighted towards the center and a size one larger than the pixel configuration, ensuring equal sums of coefficients for pixels in identical positional relationships, thereby reducing fixed patterns while preserving image detail.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Loss of energy

If a filter with center-weighted coefficients is applied while shifting the pixel group, then noise reduction is improved, but fixed patterns become more notable and image quality deteriorates

Engineering Contradiction:
Improvenoise reductionVSAvoidimage quality
Core Design Contradiction:
Loss of energyVSReliability

Solution Approach 1:

The patent applies different processing approaches to different regions based on their characteristics. Specifically, it identifies fixed pattern regions and applies targeted correction filters only where needed, while preserving detail in other regions. This local differentiation resolves the contradiction by reducing noise in fixed pattern areas without blurring important image details elsewhere.

Inventive Principle:
Principle #3Local quality

Solution Approach 2:

The patent dynamically adjusts filter parameters based on the detected fixed pattern characteristics. By changing the filter coefficients and processing parameters according to the specific fixed pattern situation, it achieves effective noise reduction while maintaining image quality. The parameter adaptation allows the system to optimize between noise reduction and detail preservation.

Inventive Principle:
Principle #35Parameter changes

2Reliability

If a filter with uniform coefficients is applied, then fixed patterns are reduced, but image detail is lost and quality deteriorates

Engineering Contradiction:
Improvefixed pattern reductionVSAvoidimage detail
Core Design Contradiction:
ReliabilityVSLoss of information

Solution Approach 1:

The patent applies uniform coefficient filters specifically to fixed pattern regions while using different processing for other areas. This localized application ensures that fixed patterns are reduced without unnecessarily blurring important image details in regions where fixed patterns are not present.

Inventive Principle:
Principle #3Local quality

Solution Approach 2:

The patent segments the image processing into different regions: fixed pattern regions and non-fixed pattern regions. Different filter strategies are applied to each segment, with uniform coefficients used only where needed for fixed pattern reduction, thereby preserving image detail in other segments.

Inventive Principle:
Principle #1Segmentation

3Reliability

If a larger filter size (6x6 pixels) is used to match the basic array pattern, then fixed patterns are reduced, but image detail and quality deteriorate

Engineering Contradiction:
Improvefixed pattern reductionVSAvoidimage detail
Core Design Contradiction:
ReliabilityVSLoss of information

Solution Approach 1:

The patent applies the 6x6 filter specifically to regions containing fixed patterns rather than uniformly across the entire image. This localized application reduces fixed patterns while preserving image detail in areas where the large filter would otherwise cause excessive blurring.

Inventive Principle:
Principle #3Local quality

Solution Approach 2:

The patent segments the image into fixed pattern regions and non-fixed pattern regions, applying the 6x6 filter only to the former. This segmentation strategy allows effective fixed pattern reduction without sacrificing image detail in regions where the larger filter size would be detrimental.

Inventive Principle:
Principle #1Segmentation

Data Source

PatentEP2763417B1Device, method and program for image processing, recording medium, and imaging device
Publication Date: 2019.02.27 FUJIFILM CORP
  • EP2763417B1 patent drawingFigure 1A~1B
  • EP2763417B1 patent drawingFigure 2A~2B
  • EP2763417B1 patent drawingFigure 3

AI summary

A method for image processing according to one aspect of the present invention includes: a step of acquiring an image taken by imaging means including an image sensor having pixel configuration with repeating cycles of M × N (M, N: integers of 2 or more) pixels; (a) a step of setting a target pixel in the acquired image and extracting K × L (K, L: integers of M<K and N<L) pixels based on the target pixel; (b) a step of calculating a pixel value of the target pixel with an operation with use of a filter which has a K × L filter size and which has specified filter coefficients arrayed therein; and a step of repeatedly executing the step (a) and the step (b) while moving the target pixel one pixel at a time with respect to the acquired image.