Methods and Techniques for Fire Detection

Signal, Image and Video Processing Perspectives

A. Enis Cetin, Bart Merci, Osman Günay, Behçet Ugur Töreyin, Steven Verstockt

• Publisher: Academic Press
• Year: 2016
• Print ISBN: 9780128023990
• eBook ISBN: 9780128026175
• Topic: Technology & Engineering
• Subtopic: Digital Media
• Index: Technology & Engineering

This book describes the signal, image and video processing methods and techniques for fire detection and provides a thorough and practical overview of this important subject, as a number of new methods are emerging. This book will serve as a reference for signal processing and computer vision, focusing on fire detection and methods for volume sensors. Applications covered in this book can easily be adapted to other domains, such as multi-modal object recognition in other safety and security problems, with scientific importance for fire detection, as well as video surveillance. Coverage includes: - Camera Based Techniques - Multi-modal/Multi-sensor fire analysis - Pyro-electric Infrared Sensors for Flame Detection - Large scale fire experiments - Wildfire detection from moving aerial platforms - The basics of signal, image and video processing based fire detection - The latest fire detection methods and techniques using computer vision - Non-conventional fire detectors: Fire detection using volumetric sensors - Recent large-scale fire experiments and their results - New and emerging technologies and areas for further research

Chapter 1

Introduction

Signal, image, and video processing are widely used in many security applications. It is possible to use visible-range and special purpose infrared surveillance cameras as well as pyro-infrared detectors for fire detection. This requires intelligent signal processing techniques for detection and analysis of uncontrolled fire behavior. As the number of recently proposed signal, image, and video processing-based fire detection methods increased over the last 10 years, a need for a book presenting basic principles of these methods emerged.

This book describes signal, image, and video processing methods and techniques for fire detection. The intended audience of the book is graduate students, researchers, and practitioners working on signal processing and computer vision-based techniques for fire detection. The book provides them with a thorough and practical overview of the state-of-the-art methods and techniques in this domain.

Sensors enhanced with intelligent signal and image processing capabilities may help reduce the detection time compared to the currently available sensors for both indoors and outdoors. This is due to the fact that cameras and other nonconventional fire sensors can monitor “volumes” and do not have the transport delay that the traditional “point” sensors suffer from. For example, it is possible to cover an area of 100 km² using a single pan-tilt-zoom camera placed on a hilltop for wildfire detection. Another benefit of volumetric sensor systems is that they can provide crucial information about the size and growth of the fire and the direction of smoke propagation.

During the last decades, improvements in the computational power of computers and the decreasing cost of imaging sensors made it possible to employ video-based fire detection techniques for real-time applications. In the literature, video fire detection algorithms developed for visible range cameras are higher in number as visible range cameras cost less compared to infrared (thermal) and time-of-flight cameras. In Chapter 2, state-of-the-art camera-based techniques for fire detection are presented.

Chapter 3 presents a set of methods for flame detection using a nonconventional sensor, a pyro-electric infrared (PIR) sensor, which is a low-cost sensor widely used for motion detection. The methods are based on the analysis of the flame flicker existing inherently in uncontrolled fires. The PIR sensors are commonly used for occupancy detection purposes in buildings. Utilizing techniques and methods presented in Chapter 3, they may turn into uncontrolled fire detectors as well.

Current methods and techniques used for multi-sensor fire analysis are described in Chapter 4. Methods in Chapter 4 are aimed at estimating the origin and growth of fires, rather than detecting them. Modeling fire behavior has important benefits in firefighting and mitigation, and is essential in assessing the risk of escalation. Techniques in Chapter 4 focus on multi-modal/multi-sensor analysis of fire characteristics, such as flame and smoke spread.

Surveillance cameras and PIR-based motion sensors are widely used in modern buildings. It is now possible to use them for fire and smoke detection by analyzing the video and signals that they generate. It is our hope that the methods and techniques discussed in this book will lead to safer buildings and living environments in the near future.

Sync Reading Stream

What's this?

Chapter 2

Camera-Based Techniques

Abstract

In the last decades, huge improvements in the computational power of computers and decreasing cost of imaging sensors have made it possible to use video-based fire detection techniques in real-time applications. Most video fire detection algorithms use visible range cameras because they are much cheaper than infrared (thermal) and time-of-flight cameras.

Keywords

Video fire detection

Background subtraction

Wavelet

Hidden Markov models

Smoke detection

Flame detection

Contents

Part 1 3

2.1 VFD in Visible/Visual Spectral Range 6

2.1.1 Color Detection 10

2.1.2 Moving Object Detection 11

2.1.3 Motion and Flicker Analysis Using Fourier and Wavelet Transforms 12

2.1.4 Spatial Wavelet Color Variation and Analysis 14

2.1.5 Dynamic Texture and Pattern Analysis 15

2.2 Spatiotemporal Normalized Covariance Descriptors 16

2.3 Classification Techniques 19

2.4 Evaluation of Visible Range VFD Methods 19

2.5 VFD in IR Spectral Range 22

2.6 Wildfire Smoke Detection Using Visible Range Cameras 23

2.7 Wildfire Detection from Moving Aerial Platforms 25

References 26

Part 2 31

2.8 Wildfire Detection with PTZ Cameras Using Panoramic Backgrounds 31

2.8.1 Panorama Generation 32

2.8.2 Wildfire Detection Algorithm 36

2.8.3 Experimental Results 40

2.8.4 Conclusion 42

References 45

Part 1

The first step in video fire detection (VFD) is to apply a background subtraction algorithm to extract moving regions in the video. Then the detected regions are analyzed temporally to be classified in terms of flickering characteristics. Markov models and frequency domain techniques can be used to identify if the flickering characteristics belong to flames. In the next step, spatial analysis is performed to check for the irregularities that are used to identify flames.

Another method is to extract features from the moving regions and use classifiers who are trained offline with videos of fire and false alarm sources. It is also possible to use active learning algorithms which are updated online to classify flame regions. The most important problem with visible range fire detections is the false alarms. Fire-colored moving regions can be diffi...

Table of contents

• Cover image
• Title page
• Table of Contents
• Copyright
• Biography
• Acknowledgments
• Chapter 1: Introduction
• Chapter 2: Camera-Based Techniques
• Chapter 3: Infrared Sensor-Based Flame Detection
• Chapter 4: Multisensor Fire Analysis
• Chapter 5: Conclusions
• Index
• Sync with Jellybooks