Media Similarity Detection via Vector Segmentation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing digital media matching algorithms struggle to identify similarities between digital media that have undergone transformations and modifications, and they fail to detect parts of media used in derivative or composite works.
Innovation Solution
A method that determines an input image's vector representation, compares it with stored images in a media database, and refines matches based on key points, alignment, and structural similarity, allowing for accurate identification of exact and direct matches even in transformed media.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If digital media matching algorithms rely on direct copying and reproduction of digital media, then the algorithms can quickly identify exact matches, but they fail to detect similarities in transformed and modified media
Solution Approach 1:
The patent segments the digital media into multiple features and characteristics for independent analysis. Instead of comparing media as a whole, the system breaks down images into visual features, audio into sound characteristics, and video into motion patterns. This allows the system to detect similarities even when the media has been transformed, as long as the underlying features remain recognizable.
Solution Approach 2:
The patent employs parameter transformation techniques that convert media into different representation spaces. Visual media is converted to feature vectors, audio to spectral representations, and video to motion fields. These parameter transformations enable the system to recognize media across different formats and transformations by comparing their underlying parameter representations rather than their surface appearances.
2Productivity
If digital media matching algorithms use direct copying comparison, then the comparison process is simple and fast, but it cannot detect parts of media used in derivative or composite works
Solution Approach 1:
The patent segments media comparison into multiple independent analysis tracks - visual feature extraction, audio characteristic analysis, motion pattern detection, and metadata examination. Each track processes specific aspects of the media independently, allowing the system to quickly identify potential matches while maintaining comprehensive analysis capability to detect derivative and composite works.
Solution Approach 2:
The patent adds multiple dimensions of analysis beyond simple direct copying comparison. It incorporates feature space representation, temporal dimension analysis for video, spectral dimension analysis for audio, and semantic dimension analysis through metadata. This multi-dimensional approach enables detection of derivative works that would be invisible to single-dimension comparison methods.
3Ease of operation
If copyright protection requires no formal registration, then the protection is automatic and accessible, but there is no way to monitor for infringement and assess validity of protections
Solution Approach 1:
The patent implements self-service monitoring where the system automatically performs copyright monitoring and infringement detection without requiring manual intervention. The automated media fingerprinting and comparison algorithms continuously scan databases and networks for copyrighted media, generating alerts and reports that enable copyright holders to monitor their work and pursue legal action when necessary.
Data Source
AI summary
A method relating to determining similarities between media, and systems and computer-readable media used to implement said method, includes determining an input image for comparison with a plurality of stored images in a media database; vectorising the input image to determine an input image vector representation; and comparing the input image with each of the plurality of stored images in a media database. The method further comprises outputting a set of direct matches and a set of exact matches as a detected comparison output for the input image. Further embodiments also relate to systems and methods for associating digital media with provenance information and creating verified copyright assets for attributing copyright ownership using the systems and methods disclosed herein.