Passenger behavior monitoring method based on machine vision

Abstract

The invention discloses a method for monitoring passenger behavior based on machine vision, comprising the steps of: 1) collecting video images of passengers riding an escalator; 2) extracting deformable component model DPM features to describe passengers' faces; 3) using SVM to walk the escalator Passenger face detection; 4) Use KCF to track the face of the escalator passenger; 5) Use an improved nearest neighbor matching method to match the face detection and tracking results; 6) Establish a face target detection intensity model, correct the face Face detection results; 7) Monitor the behavior of passengers in the escalator by analyzing the passenger detection intensity, speed and speed direction. The method of the invention can detect and track the facial objects of passengers taking the escalator in a real-time, stable and accurate manner, and conduct behavior monitoring based on the facial object information.

Description

technical field [0001] The invention relates to the technical field of image processing and behavior recognition, in particular to a method for monitoring passenger behavior based on machine vision. Background technique [0002] The intelligent video surveillance system has obvious advantages such as all-weather reliable monitoring, high monitoring accuracy, fast response and processing speed, and multi-effective use of video resources. As the security issues in public places have become more and more prominent in recent years, the call for building a safe city has become higher and higher. The intelligent video surveillance system with various obvious advantages has attracted more and more attention from government departments and urban residents. Therefore, the application of intelligent video It is of great significance for the monitoring system to monitor the behavior of moving targets accurately and stably, and to discover safety problems in time, attracting many schola...

AI Technical Summary

Problems solved by technology

[0003] So far, domestic passenger behavior monitoring methods for escalator elevator application scenarios are mostly manual monitoring methods, but this method has the disadvantages of high labor costs and unstable monitoring effects. Therefore, a passenger behavior monitoring method based on machine vision is invented. , monitor and collect images of the escalator area in real time through the camera installed directly above the upper floor plate of the escalator, and use the trained support vector machine (Support Vector Machine, SVM) face classifier to accurately detect in real time from the input image Passenger faces in the escalator area, and then through Kernelized Correlation Filter (KCF) face tracking and an improved nearest neighbor matching method to match and update the detection intensity of the face target, correct the face detection results, and finally, Monitor passenger behavior based on face targets, and realize monitoring tasks such as passenger flow statistics, congestion detection, and abnormal motion detection

Images