摘要 |
A method for fingerprinting video comprising identifying motion in a video as a function of time; using the identified motion to create a motion fingerprint; identifying peaks and/or troughs in the motion fingerprint, and using these to create a reduced size points of interest motion fingerprint. Reduced size fingerprints for a plurality of known videos can be prepared and stored for later comparison with reduced size fingerprints for unknown videos, thereby providing a mechanism for identifying the unknown videos. |