Image Comparison Using Logical Match and Blob-Based Materiality
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current image comparison methods fail to accurately identify perceptually similar images with minor differences that do not materially affect the substantive information displayed, often highlighting insignificant variations as significant, and lack the ability to account for various types of alterations.
Innovation Solution
A system and method for comparing images by calculating a logical match percentage that emphasizes large differences while negating small ones, using grayscale conversion, blob detection, and cluster analysis to identify significant image differences.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If raw data processing is used to determine image similarity by counting different pixels, then the comparison can be performed quickly, but perceptually similar images with minor differences are incorrectly evaluated as very different
Solution Approach 1:
The patent segments the image comparison process into multiple stages: first converting images to grayscale and normalizing them, then using blob detection to identify significant features, followed by cluster analysis to group similar regions, and finally calculating differences only in unmatched areas. This segmentation allows the system to process images quickly while filtering out insignificant pixel differences through hierarchical analysis.
Solution Approach 2:
The patent applies different processing qualities to different regions of images based on their importance. By using blob detection to identify significant features (such as text, graphics, or objects) and focusing comparison on unmatched areas only, the system applies higher measurement precision locally to critical regions while using faster processing for the overall comparison, thus resolving the contradiction between speed and accuracy.
2Difficulty of detecting and measuring
If edge detection, corner detection, or blob detection is used to identify specific differences, then certain types of alterations can be detected, but a comprehensive variety of alterations cannot be accounted for
Solution Approach 1:
The patent creates a universal image comparison system that can detect multiple types of alterations (text changes, graphic modifications, object repositioning, color variations) by combining several detection techniques into a single integrated framework. The cluster analysis component serves as a universal mechanism that can identify and group different types of unmatched regions, making the system adaptable to various alteration types without requiring separate specialized detectors for each.
3Measurement precision
If known image comparison techniques are used to calculate difference between images, then any minute differences can be enhanced and identified, but small differences that do not materially affect the image are incorrectly equated to large differences
Solution Approach 1:
The patent introduces cluster analysis as an intermediary mechanism between raw pixel difference calculation and final difference evaluation. This intermediary process groups pixel differences into clusters based on their spatial and visual characteristics, allowing the system to distinguish between isolated minute differences (which may be insignificant) and clustered differences (which may indicate material changes). The intermediary filtering step removes false positives while preserving true differences.
Solution Approach 2:
The patent changes the parameters used for difference calculation by first converting images to grayscale and normalizing pixel values, then using weighted calculations that consider the context and importance of different regions. This parameter transformation allows the system to maintain high sensitivity for difference detection while filtering out insignificant variations through the normalized and weighted comparison framework.
Data Source
AI summary
Methods, system, and media for comparing a set of images to determine the existence and location of any differences between the image set. The differences may be located using image comparison techniques such as SURF and Blob Detection, as well as through techniques used to identify areas of data sliding and match probabilities. A logical match probability, as well as a physical match probability, may be included in an output report with a result image highlighting the differences between the comparison images in the image set.