What is the VideoMatching?
VideoMatching library is an implementation of algorithm for seeking of shorter query video sequences in longer video streams based on video fingerprints. It it can be utilized as a basic part of solution for tasks where it is required:
- fast seeking of a query video sequence in a long video record
- storage capacity saving
- robustness and reliability
- various video-codecs support independent of codecs installed in the target system
- simple solution, easy to use with short time to market
How does it work?
Both the reference video and the query video are firstly preprocessed to extract important video features - video flags. VideoMatching library supports various codecs and formats of input video files independent of those which are currently installed in the computer. Video flags can be efectively stored on disk or in a database according to their size which is approgimately 54-times lower then size of the source video. Presence of the query video in the reference video is than evaluated by matching between selected fingerprints.
Library and algorithm key features:
- lightning fast video preprocessing - about 30x realitme
- low storage requirements - compression ration more than 1:54
- vast range of video formats supported
- suitable for database automated systems
- available as a library or DLL also with managed code support (Linux, C, Windows, C, C++, C#, .NET ...)
Algorithm is insensitive to:
- video frame rate
- video size
- brightness & color transformation
- noise in video
- letter box video format
How can you get more information?
If you have any question about our algorithms, solutions or library license please do not hesitate and contact us.