Flip-Resistant Media Fingerprinting via Nested Hash Segmentation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional media processing methods produce hashes that are dependent on the orientation of media, making it difficult to identify near-duplicate media when the media is flipped, as flipping can prevent hash matching between original and flipped versions.
Innovation Solution
A system and method for generating flip-resistant media fingerprints that includes a first hash generation component to create sets of hashes for different orientations of media, a hash augmentation component to associate index values with hash values, and a second hash generation component to compute a final hash, resulting in a fingerprint that is independent of media orientation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If conventional hash generation is used for media fingerprinting, then the fingerprinting process is simple, but the hash matching fails when media is flipped vertically or horizontally
Solution Approach 1:
The patent segments the media into multiple blocks and generates separate hash values for each block. This segmentation allows the system to process different orientations independently and combine them, resolving the contradiction between simple hashing and flip-resistant matching by dividing the media into manageable segments that can be recombined in different orientations.
Solution Approach 2:
The patent implements nested hashing where a final hash is generated by combining multiple block hashes. The structure nests block-level hashes within orientation-specific hashes, which are then nested within the final flip-resistant hash. This nested structure enables the system to maintain simplicity at each level while achieving robust flip-resistant matching at the final level.
2Adaptability or versatility
If orientation-dependent hashing is used, then the fingerprinting process is computationally efficient, but it cannot identify near-duplicate media across different orientations
Solution Approach 1:
The patent introduces dynamic hash generation where the same media block can produce different hash values depending on its orientation. The system dynamically selects and combines hash values from different orientations (original, flipped, rotated) based on the media being processed, enabling adaptability across orientations while maintaining computational efficiency through reuse of block-level hashes.
Solution Approach 2:
The patent performs preliminary hash generation for all possible orientations of media blocks in advance. By pre-computing hash values for original and flipped versions of each block, the system eliminates the need for reprocessing during matching operations, thus achieving orientation independence without sacrificing processing speed during actual fingerprinting and matching tasks.
3Reliability
If multiple hashes are generated for different orientations, then flip-resistant matching is achieved, but the computational complexity increases
Solution Approach 1:
The patent merges multiple orientation-specific hash values into a single final hash by combining them through a standardized process. Instead of maintaining separate hash structures for different orientations, the system merges block hashes from original and flipped versions into unified orientation-specific hashes, which are then merged into the final flip-resistant fingerprint, reducing overall system complexity.
Solution Approach 2:
The patent changes the parameter of hash combination by using a standardized merging function that processes hash values from different orientations uniformly. By applying consistent parameter transformation and combination rules, the system achieves flip-resistant matching without requiring complex orientation-specific processing logic, thus reducing computational complexity while maintaining reliability.
Data Source
AI summary
Systems and methods for facilitating media fingerprinting are provided. In one aspect, a system can include: a memory, a microprocessor, a communication component that receives media; and a media fingerprinting component that fingerprints the media. The media fingerprinting component employs a fingerprint generation component stored in the memory and includes: a first hash generation component that generates sets of hashes corresponding to versions of the media; and a second hash generation component that computes a final hash based, at least, on hashing the sets of hashes. In some aspects, the media fingerprinting component can generate a flip-resistant fingerprint based, at least, on the final hash. In some aspects, the flip-resistant fingerprint is the final hash.