Research On Active Object Data Association Mining Technology Based on Surveillance Video

摘要

with the rapid development of industrialization and urbanization, the public order situation becomes more and more complicated. Video surveillance is a large number of operations used to obtain characteristics of the suspect. In gang crime, the relationship between the suspects is the key clue to solving cases, which is of great significance to the detection of cases. Similar path to get active object is the relationship of the most effective method by measuring the similarity of the object activity measurement path relationship between objects, the scope for activity area is small, the path the whole situation. However, in the surveillance video, criminal gangs and sometimes appear together, sometimes separately has led to the track do not yet have similar characteristics. To solve this problem, this paper statistics on the active object in the intersection when the size of the airspace to measure the relationship between them. Firstly, we study tagging technology, which the video is converted to objects, time and space structured data. Secondly, Impact of research activities in space-time domain model objects, given time and space of the domain based on the support and confidence calculation. Finally, the data mining algorithm is given. This research study to explore spatial and temporal activities of law has significance for video annotation, and association mining technology to provide new technical means, in the criminal investigation has important application value.