Multi-Sensor Surveillance System Based on Integrated Video Analytics
- 30 November 2021
- journal article
- research article
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Sensors Journal
- Vol. 22 (11) , 10207-10222
Abstract
Image-based sensing is a powerful source of information for a wide range of applications such as surveillance, monitoring, and automation. The volume of data collected by surveillance devices is enormous; therefore, images are processed to extract application-driven data, moreover, centralized processing is unfeasible when several cameras are distributed across broad geographical regions. In such scenario, with processing limitations and bandwidth availability being important restrictions, an automated process is required for extracting meaningful information from the collected data during surveillance process and reducing operator fatigue. The proposed design, which continuously processes the image stream to meet real-time requirements, provides integrated video analytics with enhancement, panoramic views, and automated motion detection. This system enhances pictorial information for human interpretation and analyzes scene data using machine-vision modules. These features aid in the use of surveillance system for continuous monitoring of border areas without the need for human intervention under adverse-weather conditions. The proposed system architecture is based on distributed video-analytics using visual and infrared-band sensors. The system has a visual detection range of 5 km with 80% detection probability for a 2.3 m X 2.3 m target under standard clear visibility conditions. The preliminary results indicate that the system is capable of processing up to 25 video frames per second. Additionally, the implemented system records data for post-analysis and uses video analytics to reduce the bandwidth requirement of communication link. Finally, this system can be distinguished because it is flexible, compact, affordable, and scalable, and therefore, can be considered for monitoring and surveillance.Keywords
This publication has 77 references indexed in Scilit:
- A Benchmark Dataset for Outdoor Foreground/Background Extraction2013
- Phase Correlation Based Image Alignment with Subpixel AccuracyLecture Notes in Computer Science, 2013
- A two-step approach to see-through bad weather for surveillance video quality enhancementMachine Vision and Applications, 2012
- Video Surveillance Online Repository (ViSOR): an integrated frameworkMultimedia Tools and Applications, 2009
- Video Surveillance2009
- Speeded-Up Robust Features (SURF)Computer Vision and Image Understanding, 2008
- A new motion detection algorithm based on Σ–Δ background estimationPattern Recognition Letters, 2007
- New metric for predicting target acquisition performanceOptical Engineering, 2004
- Target identification performance as a function of low spatial frequency image contentOptical Engineering, 2000
- Pfinder: real-time tracking of the human bodyIEEE Transactions on Pattern Analysis and Machine Intelligence, 1997