Surveillance video abnormal event detection method based on multiple examples and time series

Abstract

The invention relates to a monitoring video abnormal event detection method based on multiple examples and time series. In the feature extraction stage, the concept of multi-instance division is used to treat each image frame in the video stream as an image package, and each package is divided into multiple disjoint examples, and then the examples belonging to the same area are divided into time order reassembled together to form the corresponding example stream. In the feature modeling stage, the continuous video frames in the sampling block are modeled using time series, and the subsequent motion trend is predicted to obtain the corresponding prediction interval, and then the overlapping prediction intervals are merged until all intervals are mutually correlated. not intersect. Finally, the abnormal event is judged according to the relationship between the actual value and each prediction interval. The present invention reduces the time complexity of detection under the premise of ensuring accuracy.

Description

technical field [0001] The invention relates to a monitoring video abnormal event detection method, in particular to a monitoring video abnormal event detection method based on multi-instance division and time series prediction. Background technique [0002] With the development of computer science and technology, the use of image processing, computer vision, machine learning and other technologies can break through the limitations of traditional video surveillance systems, and realize intelligent video analysis of video surveillance systems and active detection and real-time warning of abnormal events. Video surveillance applications in the security field are of great value. [0003] The abnormal event detection method in surveillance video is mainly divided into four basic steps: image preprocessing, basic event representation, building an abnormal detection model and judging abnormal events. Among them, the basic event representation is mainly divided into event represen...

AI Technical Summary

Problems solved by technology

[0006] 1. The algorithm extracts the average optical flow value of the image in the motion feature stage as the motion feature of the corresponding sampling block, which makes the algorithm vulnerable to the inconsistency of the optical flow feature values ​​of the same motion in different areas. The same moving object in the video, its When the distance from the camera is different, there are obvious differences between the optical flow values, which will affect the final detection;

[0007] 2. The algorithm only performs two-dimensional sampling on the video when sampling, without considering the relevant information in time, and the motion in the general video often has a front-back correlation, which will affect the detection of the model

Images