3D Wavelet Video Fingerprinting for Duplicate Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Video sharing sites face challenges in managing large video libraries, including identifying and removing unauthorized copies of copyrighted content, dealing with multiple copies of the same videos, and organizing content effectively, as existing methods fail to accurately detect similarities due to variations in compression, resolution, and timing.
Innovation Solution
A method and system that generate and compare video fingerprints representing spatial and sequential characteristics, using a three-dimensional transform to create a compact representation of video content, allowing for efficient identification and organization of videos, even with minor variations, and automatic detection of duplicates and copyrighted material.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If hash-based fingerprint functions are used to identify exact copies of video files, then identification speed is improved, but the ability to detect videos with small variations (compression, resolution, timing differences) deteriorates
Solution Approach 1:
The video is divided into multiple sequential segments, and a fingerprint is generated for each segment rather than treating the entire video as a single unit. This segmentation allows the system to compare specific portions of videos independently, making it more robust to variations in compression, resolution, and timing while maintaining efficient comparison speed.
Solution Approach 2:
The patent transitions from traditional hash-based fingerprinting to a three-dimensional wavelet transform that captures spatial and temporal characteristics across multiple scales. This dimensional transformation enables the fingerprint to represent video content in a way that is invariant to common variations, improving detection reliability while maintaining computational efficiency.
2Device complexity
If individual video frames are treated as separate images for comparison, then processing complexity is reduced, but sequential information between frames is lost
Solution Approach 1:
The patent combines spatial information from individual frames with temporal information between frames into a unified three-dimensional wavelet representation. This merging preserves sequential information while maintaining manageable processing complexity by using efficient wavelet transform algorithms that operate on the combined spatiotemporal data.
Solution Approach 2:
The fingerprint structure combines multiple types of information (spatial characteristics, temporal characteristics, and their interactions) into a composite representation. This composite approach ensures that neither spatial nor temporal information is lost, while the structured composition keeps processing complexity manageable through systematic decomposition and reconstruction.
3Measurement precision
If fingerprints are computed for the whole video file, then identification accuracy is improved, but processing time and computational resources increase
Solution Approach 1:
The video file is divided into multiple segments for fingerprint computation. This segmentation reduces the computational burden on each individual processing unit while maintaining overall identification accuracy through the systematic combination of segment fingerprints. The segmented approach enables parallel processing and reduces memory requirements.
Solution Approach 2:
Rather than computing a single comprehensive fingerprint for the entire video, the system computes multiple partial fingerprints for segments. This partial action approach reduces processing time and resource requirements while the aggregation of segment fingerprints provides sufficient accuracy for identifying duplicate and similar videos.
Data Source
AI summary
A method and system generates and compares fingerprints for videos in a video library. The video fingerprints provide a compact representation of the spatial and sequential characteristics of the video that can be used to quickly and efficiently identify video content. Because the fingerprints are based on spatial and sequential characteristics rather than exact bit sequences, visual content of videos can be effectively compared even when there are small differences between the videos in compression factors, source resolutions, start and stop times, frame rates, and so on. Comparison of video fingerprints can be used, for example, to search for and remove copyright protected videos from a video library. Further, duplicate videos can be detected and discarded in order to preserve storage space.