A Robust Abnormal Behavior Detection Method

Abstract

The invention provides a method for detecting an abnormal behavior of robustness. The method comprises the following steps of (a) extracting motion prospect information from continuous video frames by utilizing a Gauss background modeling method; (b) detecting positions of persons in the prospect by utilizing a Cascade detection method; (c) partitioning the video into a plurality of space areas, and applying the step (b) to each partitioned space area at each sampling time point to obtain a local change curve reflecting whether the persons appear or not in each space at unit time; (d) extracting behavior characteristics; (e) feeding the obtained all behavior characteristics, as training data, of the step (d) into a support vector machine to learn to obtain a multi-class classifier; (f) carrying out classification identification on video segments by utilizing the trained classifier so as to judge whether the abnormal behavior exists in the video or not. According to the method for detecting the abnormal behavior of the robustness, which is disclosed by the invention, the interference on behavior identification caused by a great amount of external factors under the real scenes can be overcome, and the exact identification on the abnormal behavior is realized.

Description

technical field [0001] The invention relates to the field of pattern recognition and intelligent monitoring, in particular to a robust abnormal behavior detection method. Background technique [0002] With the development of hardware technology, while the performance of hardware equipment is improved, the price can also be accepted by ordinary consumers, which makes monitoring equipment enter thousands of households from the public domain, greatly improving public and family safety. The need for computer technology in surveillance areas such as public safety and home security directly drives the development of this computer vision technology, which is also an important part of the development of artificial intelligence technology. In recent years, with the improvement of tracking and detection technology, the advantages of some computer vision techniques have been reflected in practical applications. However, with the rapid increase in the number of monitoring equipment, it...

AI Technical Summary

Problems solved by technology

However, with the rapid increase in the number of monitoring equipment, it has brought another problem: a large amount of data can no longer be controlled by manpower alone. Therefore, people need to use computer technology to help them automatically detect the content of interest, so as to Take follow-up actions such as real-time alarm for abnormal content received

Although there have been many studies on behavior recognition methods in recent years, and some recognition methods can achieve high recognition rates in simple scenes, in real scenes, due to a large number of factors such as illumination changes, occlusions, and viewing angle changes, making These existing behavior recognition methods are greatly limited in practical applications

Images