| 摘要 |
<p>Proposed is a video content auditing system based on a perception knowledge base, comprising: a video sample database, a sample data perception feature extraction module, a video perception knowledge base, a module for fuzzily matching perception features, a module for fuzzily distinguishing video data types, a user video data type library, and a comprehensive query module for auditing results which queries a video frame to be audited in the user video data type library and a specific auditing category to which the video frame is subordinate according to a user instruction; and/or according to the alarm setting of a video of a specific type designated by the user, sends an alarm to the user if the video to be audited contains a frame conforming to the alarm setting of the video of a specific type. Also proposed is a video content auditing method based on a perception knowledge base. The present invention automatically extracts perception knowledge from a sample video data and manages same with a database, automatically audits a video content by utilizing perception knowledge, and achieves the purpose of fully automatic audit, so that manual operations are not needed, and compilation of a specific auditing algorithm is not needed, thereby having high universality.</p> |
