Audio and video fingerprint identification method and tampering prevention system based on audio and video fingerprint streaming media

Abstract

The invention discloses an audio and video fingerprint identification method, comprising the following steps of: respectively acquiring audio and video files of a master slice and a sample slice in streaming form, and respectively extracting a master slice fingerprint feature value and a sample slice fingerprint feature value; and then, performing streaming form analysis to compare the master slice fingerprint feature value with the sample slice fingerprint feature value. The invention also discloses a tampering prevention system based on audio and video fingerprint streaming media, since small-size image data is used for extraction of video fingerprint features, complexity of extraction of the video fingerprint features is reduced, and the system has the characteristics that extraction speed is fast, storage space of fingerprint files formed by the extracted video fingerprint features is mall, and so on; indexing and retrieval of the fingerprint files are facilitated; through a hash code set combined with audio and video, a one-to-one corresponding relation is built according to actual structure of a video clip, and the tampering prevention system is built; and by comparison of the hash codes, whether frame information is tampered can be judged directly, and a tampered position can be located quickly, thus, tampering detection capability is strong, and safety is high.

Description

technical field [0001] The invention relates to the technical field of communication information security, in particular to an audio and video fingerprint identification method, and an anti-tampering system based on audio and video fingerprint streaming media. Background technique [0002] In recent years, incidents of hostile forces and illegal elements maliciously interfering with and destroying radio and television signals and radio and television facilities have occurred from time to time. Since man-made malicious sabotage is often political and concealed, we must be vigilant and take strict precautions. Many incidents have proved that once a broadcast safety accident occurs, tens of millions of people will be affected, and it will cause huge economic losses and negative social impact. [0003] At present, in the field of communication information security technology, there are many fingerprint extraction methods based on video frames, but there is no comprehensive tampe...

AI Technical Summary

Problems solved by technology

For the real-time audio and video fingerprint matching scene of this feature, there are three problems in the usual method: (1) it takes a long time to collect audio and video, because in the existing audio and video matching algorithm, although the fingerprint feature extraction algorithm is more different, But in order to obtain accurate matching results, the time of collecting audio and video (T1 and T2) needs to reach a sufficient length. Different sound scenes (music, sports, movies, talk shows, etc.), and the noise level of the environment will affect T1 There are different requirements for the time length of T2. The greater the noise, the longer the required time length. If you want to have a better audio and video recognition effect in any environment, you need to spend a long enough time to capture the master film Audio and video of audio and video samples

(2) Fingerprint matching takes a long time, because a relatively long audio and video is collected, correspondingly more fingerprint feature values ​​will be proposed, so more time is consumed when executing the audio and video fingerprint matching algorithm

(3) There may be no match that should be matched, because the network live stream will have a delay problem such as a 3s delay. In this case, the 3s master and sample audio and video are also collected, but the result is mismatched.

Images