A Surveillance Video Oriented Correlation Visual Analysis Method

Abstract

The invention discloses an associated visible analysis method facing monitoring videos. The associated visible analysis method facing the monitoring videos comprises the following steps of (1) respectively extracting a monitoring scene and monitoring object information from every monitoring video; (2) calculating similarity of a monitoring object, obtaining a matching mapping relationship of same monitoring object in different monitoring videos and establishing a video content structure; (3) establishing a scene matrix according to the video content structure and recording the appearance times of monitoring objects in different monitoring scenes; (4) finding a plurality of monitoring objects or monitoring scenes with the highest appearance rate in the object scene matrix according to the given time period condition or scene condition, then selecting a monitoring object and matching and inquiring the monitoring object with the monitoring object in the video content structure to return a plurality of similar monitoring objects. The associated visible analysis method facing the monitoring videos can excavate potential incidence relations between the different monitoring objects in the monitoring video and assists the user to analyze and find the monitoring videos.

Description

technical field [0001] The invention belongs to the technical field of computer applications, and in particular relates to a monitoring video-oriented correlation visual analysis method. Background technique [0002] With the development of science and technology, video surveillance technology has also developed rapidly. At the same time, the information of video surveillance images has also increased massively, beyond the effective processing range of manpower. It consumes a lot of time, manpower and material resources. How to effectively filter out redundant information on surveillance video content, quickly search, locate, and analyze correlations has become a hot research issue in the current computer field. A large number of scientific research and application developers have focused on surveillance video The application of visual analysis has been researched and developed. One of the methods is to preprocess the surveillance video, analyze the surveillance objects and...

AI Technical Summary

Problems solved by technology

[0003] However, there are still a lot of problems in solving massive surveillance video content search and analysis problems in this way.

First of all, since this method does not further process the monitored objects, it only displays the identified monitored objects according to the relationship of time or space, and the number of monitored objects identified from the massive surveillance videos is still far away. Far beyond the scope of human processing, it is still very difficult to find key objects and content; secondly, the process of analyzing surveillance video often needs to search back and forth according to different objects and content, rather than simply viewing and browsing. A certain correlation analysis of objects will improve the efficiency of the entire analysis process; finally, using static video summaries for display can only display video content from one dimension, and cannot fully achieve the purpose of displaying time and space by disrupting and reorganizing. Facilitate in-depth analysis of surveillance video content

Images