In order to strengthen the management and control of special crowd and increase the application effectiveness ofvideo as a typical unstructured data, this paper focuses on how to effectively extract and utilize the rich contentcontained in it to transform into intelligence information and build an effective personnel-specific tracking system. Theexperimental results show that the combination of video motion region extraction and personnel image features can moreaccurately complete person detection, and the introduction of multi-feature fusion to extract special person regions, suchas clothing logo identification and other areas, for specific person to carry out feature labeling to index, improve theaccuracy of personnel tracking. From the perspective of police technology innovation, this paper introduces the methodof personnel characteristics analysis, comparison, retrieval, data mining and intelligent early warning to improve theapplication level of valuable personnel characteristics data in video images, and establish a feature analysis system toachieve continuous tracking of targets.
展开▼
