Image Hash Framing for Privacy-Preserving Illegal Image Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing image analysis technologies struggle to accurately identify child abuse images while maintaining data privacy, as conventional hash values are sensitive to pixel changes and can be easily circumvented by image manipulation, leading to false positives and non-compliance with privacy regulations.
Innovation Solution
A method involving creating multiple frames around suspected areas of an image, assigning each frame a hash value and a reference point, and comparing these values at a service provider site to determine if a specified ratio of matches is met, ensuring accurate identification without revealing the image content.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional hash value comparison is used to detect illegal images, then image identity can be accurately identified, but any single pixel change causes the hash value to change completely making detection ineffective
Solution Approach 1:
The patent divides the image into multiple frames and generates separate hash values for each frame. Instead of using a single hash value for the entire image, the system creates a collection of frame-level hash values that collectively represent the image content. This segmentation approach ensures that minor pixel changes in one area do not affect all hash values, thereby improving detection robustness while maintaining identification accuracy.
2Reliability
If perceptual hashing is used to reduce sensitivity to pixel changes, then detection robustness improves, but image size changes cause the hash value to change leading to false negatives
Solution Approach 1:
The patent performs preliminary actions by dividing the image into multiple frames and generating hash values for each frame before comparison. This pre-processing step creates a structured representation that is more resilient to both pixel changes and size variations. By establishing this frame-based hash structure in advance, the system can accurately match images even when they undergo minor transformations.
3Measurement precision
If service providers analyze image content to identify illegal content, then detection accuracy improves, but data privacy regulations are violated
Solution Approach 1:
The patent extracts only the essential features needed for detection by generating hash values from divided frames, rather than analyzing the complete image content. This extraction approach allows service providers to perform detection operations on compressed representations (hash values) instead of original images, thereby maintaining detection accuracy while preserving user privacy and complying with data protection regulations.
4Productivity
If perceptual hash value is used for detection, then processing efficiency improves, but false positives increase causing false reports to authorities
Solution Approach 1:
The patent merges multiple frame-level hash values into a collective representation for final comparison. By combining the results from multiple frames rather than relying on a single perceptual hash value, the system reduces false positives while maintaining processing efficiency. The merged evaluation provides more reliable detection results, ensuring that only genuine matches are reported to authorities.
Data Source
AI summary
One example method includes receiving a digital image, receiving an input list that includes a respective hash for each frame in a group of frames, creating new frames on a particular area of the digital image, and the new frames are created and located based on the information in the input list, obtaining a respective hash for content included within each of the new frames, comparing one of the hashes generated for one of the new frames with a hash from the input list and, when the hashes match, continue the comparing for all frames of the particular area, and when all hashes from the input list have been checked, determining whether or not the digital image is an illegal image.