Image Noise Intensity Estimation via Sub-block Error Calculation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing image noise estimation methods face challenges in accurately estimating noise intensity, particularly in complex systems where noise does not conform to a certain probability distribution, and are affected by texture and edge information, leading to reduced accuracy.
Innovation Solution
An image noise intensity estimation method that filters a first image to obtain a second image, divides both into sub-blocks, calculates error values between corresponding sub-blocks, and estimates noise intensity using these error values, thereby avoiding the impact of motion factors and improving estimation accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If noise model-based estimation method is used, then the method can establish a probability model for image noise, but the estimation accuracy differs greatly from real values when noise does not conform to the probability distribution
Solution Approach 1:
The patent changes the estimation approach from model-based probability parameters to direct pixel intensity difference parameters. By calculating the absolute difference between original and filtered pixel values, the method adapts to any noise distribution without requiring prior knowledge of noise characteristics, thereby improving both adaptability and measurement precision simultaneously
2Measurement precision
If transform domain-based method is used, then the method can transform the image to estimate noise intensity, but the noise characteristics are greatly affected by texture and edge characteristics
Solution Approach 1:
The patent extracts only the noise component by calculating the difference between the original image and the filtered image. This extraction method isolates noise from texture and edge information, as the filtering process preserves structural features while removing noise, allowing accurate noise estimation without interference from texture and edge characteristics
3Measurement precision
If image sub-block-based method with motion detection is used, then the method can screen out image still region, but the accuracy of motion detection is difficult to ensure and calculation complexity increases
Solution Approach 1:
The patent uses the filtered image itself to estimate noise intensity by calculating pixel differences, eliminating the need for external motion detection modules. The filtering process automatically provides a denoised version of the same frame, allowing noise estimation without requiring motion compensation or inter-frame analysis, thereby reducing system complexity while maintaining accuracy
Data Source
AI summary
An image noise intensity estimation method, an image noise intensity estimation device, and an image recognition device are disclosed. The method includes: obtaining a first image to be estimated; filtering the first image to obtain a second image; dividing the first and second images to obtain a plurality of first image sub-blocks and a plurality of second image sub-blocks respectively; calculating error values between the first image sub-blocks and the second image sub-blocks in corresponding positions; and estimating the noise intensity of the first image according to a plurality of error values obtained by calculation. The method can improve the accuracy of noise estimation.