Dynamic Thresholding for Image Cross Correlation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing image cross-correlation methods face challenges in accurately locating multiple occurrences of an item of interest within a document or image due to the need for setting a static threshold, which can result in either too few meaningful results or an overwhelming number of false results if the threshold is set too high or low.
Innovation Solution
A dynamic thresholding method is implemented, where a median absolute deviation analysis is used to determine a threshold value for each image, allowing for precise location of items of interest by identifying outliers and focusing on meaningful results.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of operation
If a static predefined threshold is used for cross correlation, then the method is simple and easy to implement, but the accuracy of locating items of interest deteriorates due to either too few meaningful results or too many false results
Solution Approach 1:
The patent applies dynamics by transitioning from a static predefined threshold to a dynamic threshold that adapts to each document's characteristics. The system calculates a unique threshold for each document based on its specific content and properties, allowing the threshold to respond to varying document conditions rather than remaining fixed across all documents.
Solution Approach 2:
The patent implements parameter changes by modifying the threshold parameter based on document-specific analysis. Instead of using a constant threshold value, the system adjusts the threshold parameter dynamically according to the measured characteristics of each document, such as content density and item distribution patterns.
2Reliability
If a high threshold is set to reduce false results, then the number of false positives decreases, but the quantity of meaningful results located also decreases
Solution Approach 1:
The patent changes the threshold parameter dynamically based on document characteristics. By analyzing each document's specific properties such as content density and item distribution, the system adjusts the threshold to an optimal level that maintains high reliability while preserving the maximum number of meaningful results.
Solution Approach 2:
The system employs feedback by using document-specific analysis results to inform threshold selection. The threshold is determined based on feedback from analyzing the document's content characteristics, item distribution patterns, and statistical properties, creating a closed-loop system that optimizes both reliability and result quantity.
3Quantity of substance
If a low threshold is set to capture more results, then the quantity of located items increases, but the number of false results increases and obfuscates desired results
Solution Approach 1:
The patent dynamically adjusts the threshold parameter based on document-specific analysis, finding the optimal balance point for each document. This prevents the threshold from being too low (which would capture false results) or too high (which would miss meaningful results), adapting to each document's unique characteristics.
Solution Approach 2:
The patent applies local quality by tailoring the threshold to each specific document's characteristics rather than using a uniform approach. Each document receives a customized threshold based on its local properties such as content density, item distribution, and statistical measures, ensuring optimal performance for each document's specific context.
4Measurement precision
If a dynamic threshold is calculated for each document, then the accuracy of locating items of interest improves, but the device complexity and computational requirements increase
Solution Approach 1:
The patent changes the threshold parameter dynamically through automated calculation based on document characteristics. The system computes statistical measures such as median and mean values from the cross-correlation results and uses these to determine the optimal threshold, replacing manual threshold setting with automated parameter adjustment.
Solution Approach 2:
The system implements self-service by automatically determining its own threshold parameter without external intervention. The patent employs algorithms that autonomously analyze document properties and calculate the appropriate threshold, allowing the system to self-adjust and optimize performance without requiring manual configuration or complex external control mechanisms.
Data Source
AI summary
The present disclosure relates to a computer-implemented system and method for finding matching occurrences of an item of interest (or image or sub-image) within a document (or larger image) via cross correlation and setting a dynamic threshold for each document (or larger image). The described system and method are capable of matching and locating the one or more items of interest within each specific document (or larger image).