Robust Video Feature Using Persistent Homology With Resistance to Various Modification

With the spread of network services, video distribution services have been expanding remarkably. On the other hand, the act of uploading other people’s works without their permission has become a problem. Unauthorized reproduction of digital contents is an extremely malicious act that damages the interests of authors, and it is desirable to be able to detect it promptly. For this purpose, it is necessary to distinguish whether video images are visually equal or not. Therefore, we proposed a robust video hash based on persistent homology. Experimental results show that the features used in this study are robust to various modification operations and can be used for video identification.