Stochastic Resonance Image Processing for Singular Portion Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing image processing methods struggle to extract singular portions, such as flaws, from images with varying lightness and hues with stable accuracy, often leading to incorrect or missed detections due to the need to tune noise strength and threshold values based on image position and gradation.
Innovation Solution
An image processing apparatus and method that adjusts noise strength and threshold values for stochastic resonance processing based on individual pixel data, using binary processing and parallel synthesis to enhance detection accuracy across different image gradations.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If a fixed noise strength and threshold value are used for stochastic resonance processing, then the processing is simple and fast, but the detection accuracy varies depending on image position and gradation
Solution Approach 1:
The patent applies dynamics by making the noise strength and threshold value adjustable and adaptive rather than fixed. The system dynamically changes these parameters based on the input image data characteristics, allowing optimal detection accuracy across different image regions and gradations while maintaining a relatively simple processing framework.
Solution Approach 2:
The patent implements parameter changes by modifying the noise strength and threshold value parameters according to the input image data. Different parameters are set for different pixel signals based on their characteristics, enabling accurate detection across varying lightness and hues without requiring complex multi-stage processing.
2Measurement precision
If noise strength is increased to improve detection of singular portions, then detection sensitivity improves, but false positives increase in regions with varying lightness and hues
Solution Approach 1:
The patent applies local quality by setting different noise strength and threshold value parameters for different pixel signals based on their local characteristics (lightness and hue). This allows each region of the image to be processed with optimally tuned parameters, improving singular portion detection accuracy while preventing false positives in regions with varying properties.
Solution Approach 2:
The patent implements parameter changes by adjusting noise strength and threshold values according to the specific characteristics of each pixel signal. This dynamic parameter adjustment ensures that detection sensitivity is optimized locally for each region, maintaining both high detection accuracy and reliability across the entire image.
3Measurement precision
If multiple fixed parameters are used for different image regions, then detection accuracy improves, but the processing time and complexity increase
Solution Approach 1:
The patent implements parameter changes in an efficient manner by setting parameters based on the input image data characteristics without requiring extensive preprocessing or multiple processing stages. The system dynamically adjusts noise strength and threshold values for each pixel signal, achieving high detection accuracy while maintaining relatively fast processing speed.
Applied Scientific Principles
This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.
Function Achieved in This Case
Stably extracts singular portions from images with varying lightness and hues by dynamically setting noise strength and threshold values for each pixel, improving detection accuracy and reducing false positives or missed detections.
Implementation Method 1
a stochastic resonance process is useful. The stochastic resonance process is a phenomenon in which an input signal buried in noise is further added with noise and the resultant signal is subsequently subjected to nonlinear processing to thereby emphasize a detection target signal
Data Source
AI summary
An image processing apparatus a processor that acquires reading image data, composed of a plurality of pixel signals, and executes stochastic resonance processing, in which each of the plurality of pixel signals is added to noise and is subjected to a binary processing, and a plurality of results, obtained by performing the noise addition and the binary processing on the plurality of pixel signals in parallel, are synthesized. With regard to a pixel signal as a processing target among the plurality of pixel signals, at least one of a strength of the noise and a threshold value used for the binary processing is set based on a pixel signal of the input image data corresponding to the pixel signal. In addition, the processor outputs the result of the stochastic resonance processing.