.png){kind=icon}
.png){kind=icon}
Return
A hierarchical multi-modal signature method for OTT content identification and redistribution detection
.png){kind=icon}
PRE
AI
B
S
Y
DOI：10.1007/s00530-026-02565-7.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
With the rapid expansion of OTT streaming services, robust identification of redistributed video content has become increasingly important. Redistributed OTT content often undergoes transformations such as re-encoding, subtitle insertion, resolution change, aspect-ratio modification, brightness adjustment, mirroring, rotation, and frame-rate reduction. Since these transformations can change the binary and visual representation of content, conventional metadata-based and exact hash-based methods are limited in identifying transformed copies. This paper proposes a hierarchical multi-modal signature method for OTT content identification and redistribution detection. The proposed method extracts representative frames from HLS-based media segments by considering temporal position and visual variation. Visual, hash, and descriptive signatures are then generated from representative frames, media segments, metadata, and rights-related information. Content identification is performed through coarse-level candidate filtering using hash and descriptive signatures, followed by fine-level verification using visual signatures. Experimental results show that the proposed method achieved an accuracy of 96.08%, precision of 99.92%, recall of 95.80%, F1-score of 97.81%, and false positive rate of 0.80%. It also reduced the average number of candidates from 1,000 to 96 and achieved an average total query processing time of 1.4212 s. These results indicate that the proposed method provides effective identification performance and improved retrieval efficiency under various OTT redistribution conditions.
Keywords:
OTT Streaming
Content Identification
Redistribution Detection
Hierarchical Signatures
Metadata-based Verification
Journal
IF:
3.1
Papers:
2.7K
Citations:
2.7K
Organization
Cited Papers