Image Processing Apparatus Noise Intensity Estimation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing image processing methods struggle to accurately estimate noise intensity in digital images, leading to inadequate noise removal or information loss during character extraction from documents, as they often rely on generic filter settings that do not account for varying noise levels.
Innovation Solution
An image processing apparatus and method that divides an input image into multiple areas, calculates difference images before and after noise removal, computes relative pixel intensities, detects frequency distributions of these values, and estimates noise intensity based on background area distributions to select appropriate noise removal filters.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Object-affected harmful factors
If a filter with large noise removal effect is used, then noise is effectively removed, but character information may be removed as well
Solution Approach 1:
The patent applies parameter changes by adjusting filter parameters based on the detected noise intensity. The system calculates noise intensity from the frequency distribution of relative values, then selects or adjusts filter parameters accordingly - using stronger filters for high noise intensity and weaker filters for low noise intensity, thus resolving the contradiction between noise removal effectiveness and character information preservation
Solution Approach 2:
The system implements dynamics by making the filter selection adaptive rather than static. The filter parameter is dynamically adjusted based on the measured noise characteristics of the input image, allowing the noise removal strength to match the actual noise level present, preventing both over-processing and under-processing
2Loss of information
If a filter with small noise removal effect is used, then character information is preserved, but noise is not adequately removed
Solution Approach 1:
The system changes filter parameters based on measured noise intensity. By detecting the actual noise level through frequency distribution analysis and adjusting the filter strength accordingly, the system ensures adequate noise removal while preserving character information - using stronger filters when noise is high and weaker filters when noise is low
3Ease of operation
If generic filter settings are used, then processing is simple, but noise removal is inadequate for varying noise levels
Solution Approach 1:
The system applies self-service by automatically detecting noise intensity and selecting appropriate filter parameters without requiring manual intervention. The apparatus autonomously calculates the frequency distribution of relative values, determines noise intensity, and configures the filter accordingly, achieving both simplicity and precision
Solution Approach 2:
The system implements feedback by using the detected noise intensity as input to adjust filter parameters. The noise measurement feedback loop allows the system to adapt filter settings based on actual image conditions, resolving the contradiction between operational simplicity and adaptive precision
Data Source
AI summary
An image processing method comprising, dividing an input image into a plurality of divided images, calculating a difference image between a divided image before noise removal and a divided image after noise removal for each of the plurality of divided images, calculating a relative value between a pixel intensity in the divided image before noise removal and a pixel intensity in the difference image for each of the plurality of divided images, detecting a frequency distribution of relative values in a background area of the input image, contained in frequency distribution of relative values calculated with respect to the plurality of divided images, and estimating an intensity of noise in accordance with the frequency distribution of the relative values in the background area.