Multimedia Object Identification via Weighted Fingerprint Decisions
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing multimedia fingerprint algorithms face challenges in achieving high robustness and sensitivity simultaneously, making it difficult to identify multimedia objects accurately across content modifications and changes.
Innovation Solution
The method involves acquiring query fingerprints using multiple diverse fingerprint algorithms, calculating decisions using classifiers, and combining these decisions as a weighted sum to identify matching multimedia objects, with the option to train a model using training data to minimize identifying errors.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If a single fingerprint algorithm is used, then the system complexity is low, but the identification accuracy and reliability are insufficient
Solution Approach 1:
The system divides the identification task into multiple independent fingerprint algorithms (audio fingerprinting, video fingerprinting, image fingerprinting), each processing specific features separately. This segmentation allows each algorithm to specialize in detecting particular content characteristics while maintaining overall system manageability and improving cumulative identification reliability.
Solution Approach 2:
Multiple fingerprint algorithms are merged into a unified identification system where their results are combined through a decision-making process. The system integrates outputs from diverse fingerprinting methods (audio, video, image) to achieve more reliable identification than any single algorithm could provide alone, resolving the contradiction between reliability improvement and complexity management.
2Reliability
If multiple fingerprint algorithms are used, then the robustness and sensitivity of identification are improved, but the computational complexity and processing time increase
Solution Approach 1:
The system performs preliminary fingerprint extraction for each media type (audio, video, image) independently before the final identification comparison. This preliminary action prepares all necessary features in advance, allowing the actual identification process to proceed more efficiently by comparing pre-processed fingerprints rather than analyzing raw media files during the matching phase.
Solution Approach 2:
The system dynamically selects and weights different fingerprint algorithms based on the query type and available reference data. Not all algorithms are applied with equal intensity in every scenario - the system adjusts which fingerprints are generated and how they are weighted, optimizing processing time while maintaining robustness and sensitivity for each specific identification task.
3Measurement precision
If multiple fingerprint algorithms are used, then the identification accuracy is improved, but the device complexity increases
Solution Approach 1:
The complex identification task is segmented into multiple specialized fingerprint algorithms, each responsible for extracting specific features from different media types. This segmentation improves measurement precision by having dedicated algorithms for audio, video, and image analysis, while keeping each individual algorithm relatively simple and manageable.
Solution Approach 2:
The system implements a universal identification framework that can handle multiple media types (audio, video, image) through a common architecture. This multi-functional system uses the same overall structure and decision-making process across different fingerprint types, reducing the need for separate complex systems for each media type and thereby controlling device complexity while maintaining high identification accuracy.
Data Source
Figure 1~2
Figure 3
Figure 4~5
AI summary
Embodiments of identifying multimedia objects based on multimedia fingerprints are provided. Query fingerprints are derived from a multimedia object according to differing fingerprint algorithms. For each fingerprint algorithm, decisions are calculated through at least one classifier corresponding to the fingerprint algorithm based on the query fingerprint and reference fingerprints, the reference fingerprints being derived from reference multimedia objects according to the same fingerprint algorithm. Each of the decisions indicates a possibility that the query fingerprint and the reference fingerprint are not derived from the same multimedia content. For each of the reference multimedia objects, a distance is calculated as a weighted sum of the decisions relating to the reference fingerprints. The multimedia object is identified as matching the reference multimedia object with the smallest distance less than a threshold.