Image Matching for Mark Detection in Damaged Content
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for detecting unauthorized duplication of digital content, such as watermarks, are ineffective due to ease of removal and difficulty in identifying the source of leaked content.
Innovation Solution
A method and apparatus that generate matchable content by adjusting the sizes and locations of cuts in leaked content to match the original content, allowing for accurate detection of added marks using image matching techniques.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If watermark is added to content to prevent unauthorized duplication, then content protection is improved, but content visibility deteriorates and watermark removal becomes easier
Solution Approach 1:
The patent creates a digital fingerprint by extracting unique characteristics from the original content and embedding it into a copy of the content. This fingerprint serves as an invisible marker that travels with the content through distribution channels, enabling identification of leaked copies without affecting the content itself or requiring visible watermarks.
Solution Approach 2:
The patent replaces the mechanical approach of visible watermarks with a digital signal processing approach. By transforming content into the frequency domain and embedding fingerprint signals in specific frequency components, the system achieves content protection through mathematical transformations rather than visual overlays, making the protection invisible and more resistant to removal.
2Ease of operation
If traditional watermark detection is used, then detection process is simple, but detection accuracy deteriorates due to content damage during leakage
Solution Approach 1:
The patent performs preliminary processing on the suspected leaked content before comparison, including noise filtering, normalization, and feature extraction. This preliminary action prepares the data in an optimal state for accurate fingerprint matching, ensuring that detection accuracy is maintained even when the content has been damaged or modified during leakage.
Solution Approach 2:
The patent changes the parameter space by transforming both the original content and suspected leaked content into the frequency domain using Fourier transforms. By comparing fingerprints in the frequency domain rather than the spatial domain, the system achieves robustness against common leakage operations such as cropping, resizing, and compression, thereby maintaining high detection accuracy.
Data Source
AI summary
An apparatus for detecting a mark by using image matching includes a damaged content obtaining unit configured to obtain damaged content generated based on original content including a plurality of cuts; a pre-processed content generating unit configured to generate pre-processed content by merging one or more images included in the damaged content or removing a partial region of one or more cuts included in the damaged content; a matchable content generating unit configured to generate matchable content, in which sizes or locations of one or more cuts included in the pre-processed content are adjusted by comparing one or more cuts included in the original image and the one or more cuts included in the pre-processed content; a matching unit configured to compare the original content with the matchable content, for each cut.