Storage Video Authentication Using ROI Frame Similarity

The patent applies preliminary action by performing fake video detection during the storage process itself rather than after retrieval. The storage device executes detection operations on video data before it is fully stored or while it is being stored, generating detection information that is saved alongside the video data. This approach ensures data integrity is verified upfront, preventing fake video from being stored in the first place, while avoiding the need for complex post-retrieval detection systems.

The patent implements partial action by performing detection operations selectively rather than comprehensively on all video data. The detection mechanism focuses on identifying key features and patterns that indicate fake video, using algorithms that analyze specific aspects of the video data such as frame consistency, metadata validation, and codec verification. This selective approach achieves sufficient detection accuracy without requiring exhaustive analysis of every byte of video data, thereby reducing processing time and computational resource requirements.

The patent applies the extraction principle by separating detection information from the main video data storage. Instead of storing comprehensive detection results for every video file, the system extracts and stores only essential detection information such as authentication tags, checksums, or condensed verification data. This extracted detection information is sufficient to verify authenticity when needed but occupies minimal storage space compared to storing full detection datasets. The separation allows video data to be stored efficiently while maintaining verification capabilities.