Adaptive Isolated Point Noise Detection in Endoscope Imaging
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing image processing methods struggle to accurately detect and correct isolated point noise, especially in areas with local changes in pixel values, such as edges, due to the use of fixed threshold values that are not synchronized with the image's pixel value changes.
Innovation Solution
An image processing device and method that detect isolated point noise by calculating first and second index values representing the range of pixel values within a given area, excluding the minimum and maximum values, and determining the presence of noise based on these indices, allowing for adaptive detection and correction of noise without being affected by local changes in pixel values.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If fixed threshold values are used for noise detection, then the detection method is simple, but the detection accuracy deteriorates in areas with local changes in pixel values
Solution Approach 1:
The patent applies dynamics by replacing fixed threshold values with dynamic index values that adapt to local pixel value changes. The first index value represents the range of sorted pixel values, and the second index value represents the range of a subset of these values. These index values automatically adjust to local variations in the image, enabling accurate noise detection in both uniform and edge regions without requiring complex adaptive algorithms.
Solution Approach 2:
The patent changes the parameter from fixed threshold to dynamic index values derived from the actual pixel value distribution. By sorting pixel values and calculating ranges (first index value as the full range, second index value as a subset range), the system transforms static detection parameters into dynamic ones that reflect local image characteristics, resolving the contradiction between simplicity and accuracy.
2Reliability
If median filtering is applied to correct isolated point noise, then noise correction is achieved, but edge details may be lost due to overcorrection
Solution Approach 1:
The patent applies local quality by using different detection criteria for different regions based on their local pixel value characteristics. In uniform regions, the index value method effectively identifies isolated point noise. In edge regions with local pixel value changes, the dynamic index values adapt to the local variation, preventing misidentification of edge pixels as noise. This localized adaptive approach corrects noise while preserving edge details.
3Productivity
If fixed threshold values are used for noise detection, then the processing speed is fast, but the adaptability to different image regions deteriorates
Solution Approach 1:
The patent applies self-service by enabling the detection system to automatically adapt to different image regions through self-calibration. The first and second index values are calculated from the actual pixel values in each local region, allowing the system to self-adjust its detection criteria without external intervention or complex adaptive algorithms. This self-service mechanism maintains processing speed while achieving region-specific adaptability.
Data Source
AI summary
An image processing device includes an isolated point noise detection section, and an isolated point noise correction section. The isolated point noise detection section determines whether or not isolated point noise is included within a given area based on a first index value that represents the range of first to nth pixel values being obtained by arranging the pixel values of pixels within the given area including an attention pixel in ascending or descending order, and a second index value that represents the range of a pixel value group being obtained by excluding at least one of the first pixel value and the nth pixel value from the first to nth pixel values. The isolated point noise detection section determines whether or not the attention pixel is a pixel that corresponds to isolated point noise when it has been determined that isolated point noise is included within the given area.