BSE532 Fire Engineering Systems

MSc Lecture 2:
Advanced fire detection on
image processing electronics

Dr. Gigi C.H. LUI

Department of Building Environment and Energy Engineering
The Hong Kong Polytechnic University
Hong Kong, China
MScFES2022-02-AdvFD.ppt 1
Contents
1. Video image processing for fire and
smoke detection
2. Development of video fire and
smoke detection system
3. Automatic fire detection system
using video sequences
4. Image processing

2
1. Video Image Processing for
Fire and Smoke Detection
• For better fire protection
• 3 signatures
• Smoke, heat and light
• Automatic detection and alarm by video
images
• Recognition of fire and smoke.
• Integrate fire and smoke detection with
security systems [Gorovici] in buildings
• Integrated with smoke extraction system
Fire Detection System 3
• Real-time with color video input
• Advantages:
• Standard video technology and normal
human vision share similar image
recognition limits.
• effective use of smoke, heat and light as the
main fire parameters.
• Wide area detection capabilities.
• Superior in spatial analysis
• Rapid and efficient detection

Fire Detection System 4

• Visual image of area
• assess action to be taken
• make intelligent decisions automatically,
• identification of fire level
• determine type of intervention

• Trend in video image detection systems

• Software innovations
• Camera information
• Existing CCTV cameras
• Reduce cost
• Gaining efficiency
Fire Detection System 5
• 3 main techniques
• Histogram based

• Temporal based

• Rule based

Fire Detection System 6

2. Development of Video Fire and
Smoke Detection System
• CCTV-based fire detection system for power plant
• Proposed by Aird and Brown.
• Characteristics:
• Sophisticated computer analysis of video images

• Detect fire in its incipient stage

• Remotely, using inexpensive equipment,

• Raise fire alarm.

• Successfully determine the nature of the object

observed.
Fire Detection System 7
– Level of likelihood to be smoke or not.
– Filter techniques.
– Computer program to detect and locate
smoke and vapor.
– Realistic medium for early detection of
smoke
– Software approach used is right route.

Fire Detection System 8

• Techniques to reduce false alarms in forest fire
detection system based on infrared cameras
• Real-time identification of smoke images by
clustering motions on a fractal curve with a
temporal embedding method
• Smoke detection in tunnels using video images
• Advanced technologies in image processing
and pattern recognition

Fire Detection System 9

• Breakthroughs in signal processing
fundamentals
• Advances and extensions intechniques.
• signal processing techniques from
single-rate to multi-rate processing,
from time-invariant to adaptive
processing, from frequency-domain
(the traditional Fourier transform) to
time-frequency analysis, and from
linear to non-linear signal
processing.

Fire Detection System 10

– A fuzzy-logic-control-based filter for image
enhancement was studied by Farzam et al.
– Remove impulsive noise and smooth out
Gaussian noise while preserving edges and
image details efficiently
– Nonlinear operators known as Fuzzy
Inference Ruled by Else-action (FIRE)
operators processing image data based on
fuzzy reasoning was studied by Russo.

Fire Detection System 11

• A new automatic image segmentation
method was proposed by Fan et al.
• Color edges in an image
• Major geometric structures in an image
• Centroids between adjacent edge regions
for seed region growing (SRG).
• Homogeneous image regions

Fire Detection System 12

• Development of Main Hardware
Device-Video Cameras
• New kind: Day/Night Cameras
• Combine advanced performance benefits
of super dynamics with Day/Night
operation
• high quality images in virtually 24 hour-
day/night,
• all-condition level,
• any lighting condition.
• The units are IR sensitive for applications
in virtual darkness.
Fire Detection System 13
• Automatic Fire Detection System Using
Video Sequences
• Video surveillance is based on video image
detection systems (computer technology and
machine vision technology) to analyze the data
collected with video cameras.

Fire Detection System 14

 Real time Video sequence
Video Digitizer The information
camera
of color and
motion

Autonomous
segmentation by
fast entropic
thresholding,
The rapid Region growing,
detection Otsu method,
algorithm Etc. The false alarm
phenomena

Fire Detection System 15

3. Automatic Fire Detection System Using Video Sequences

• Video hardware to digitize all images in real time.

• Studying the color and motion information computed
from video sequences.
• Computer analysis  studying the rapid detection
algorithms of smoke and flame by looking
• Video information through filters
• Relationships between the filtered characteristics 
confidently predict the presence of smoke and/or flame.

Fire Detection System 16

• System can eliminate or mask off parts of the
image from detection on an individual pixel-
by-pixel basis.
• Employing highly complex algorithms to
process video information from many cameras
simultaneously

Fire Detection System 17

 Software development for Automatic Fire
Detection System Using Video Sequences
• Digital and analogue representation of
image
A B  f (0,0) f (0,1)  f (0, M  1) 
 f (1,0) f (1,1)  f (1, M  1) 
C f ( x, y )  
     
 
 f ( N  1,0) f ( N  1,1)  f ( N  1, M  1)
 [ f (i, j )] (i  0,1,  N  1; j  0,1,  , M  1)

• In the case of
• a gray-scale image,
• a color image,
• a binary image. 18
 4. Image processing
Light source
conversion changes the representation
of the image from optical light form to
a continuously varying electric signal

Light reflected
Object By object

Electrical signal Digital signal

Video
Digitizer
Camera

Digital
Optical Analog Image
Image Image Form
Form Form

Any operation that acts to improve, the manipulation and

correct, analyze, or in some way analysis of pictorial information.
change an image is called
image processing.
19
Image Enhancement
• Image Enhancement by Histogram-Modification Techniques
• Histogram
• A histogram is a graphic representation of the frequency of
occurrence of the intensity values in an image
• This histogram is a graph showing the number of pixels in an
image at each different intensity value found in that image. In
other words, it represents the relative frequency of occurrence of
the various gray levels in the image.
• The histogram of a gray levels in an image is defined by

n(i)=sum of gray levels in

n(i ) the image having the value i;
h(i ) 
n

Fire Detection System 20

 The construction of a histogram of an image is very simple.
The image is scanned in a single pass and a running count of the
number of pixels found at each intensity value is kept.
This is then used to construct a suitable histogram.

For an 8-bit grayscale image there are 256 different possible intensities,
and so the histogram will graphically display 256 numbers showing the
distribution of pixels amongst those grayscale values.

Fire Detection System 21

 For gray levels that assume discrete value,
we deal with probabilities given by the relation

where L=the number of levels

pr(rk)=the probability of the kth gray level
nk=the number of times this level appears in the image
n=total number of pixels in the image.

Fire Detection System 22

 The technique used for obtaining a uniform histogram is known as
histogram equalization or histogram linearization.

It is noted that the transformation function T(rk) can be computed directly

from the image in question by using the last formula.

Fire Detection System 23

Image Enhancement
• Image smoothing
• Filter the noise of the image
• Median smoothing filter
• Low pass filter
• Remove high spatial frequency noise from a digital image
• Image sharpening
• Differentiation and gradient

Fire Detection System 24

Segmentation
• process that subdivides an image into its
constituent parts or objects.
• One of the most important elements in
automated image analysis

Fire Detection System 25

 Edge Detection
The edge is defined as the boundary between two regions
with relatively distinct gray-level properties.
Edge detection is by far the most common approach for detecting
meaningful discontinuities in gray level.
Basically, the idea underlying most edge-detection techniques
is the computation of a local derivative operator.
Consider the sub-image area shown in the following:

x1 x2 x3
x4 x5 x6
Represent the gray
Level at x,y
x7 x8 x9
Fire Detection System 26
Define the component of the gradient vector in
the x direction as

Gx=( x7+2 x8+ x9)-( x1+ 2x2+ x3)

And in the y direction as

Gy=( x3+2 x6+ x9)-( x1+ 2x4+ x7)

Fire Detection System 27

 The Gx and Gy, as given in the previous two
formulas, can be computed by
using the kernel shown in the following

- 1 -2 -1 -1 0 1

0 0 0 -2 0 2

1 2 1 -1 0 1

The Sobel operator performs a 2-D spatial gradient measurement on an image and so
emphasizes regions of high spatial gradient that corresponds to edges.

Fire Detection System 28

• Convolving these kernels with an image f(x, y) yields the
gradient at all points in the image, the result often being referred
to as a gradient image.

• There are numerous ways to generate a gradient image based on

the use of thresholds.

Fire Detection System 29

 Thresholding
In many vision applications, it is useful to be able to separate out the regions of the
image corresponding to objects in which we are interested.

The simplest and most widely used segmentation method is thresholding

It consists of setting background values for pixels below a threshold value T

and a different set values for the foreground.
If the input image is f(x,y) and thresholded image is g(x,y), the equation
of the thresholding operator is given by:

0 if f ( x, y )  T
g ( x, y )  
1 otherwise

Fire Detection System 30

Thresholding at 100 grayscale

Fire Detection System 31

Fire image segmentation using Shannon’s entropy method

Figure 1. 24 bit true color image of town gas flame Figure 2. 256 gray level image of town gas flame

Figure 4. Distribution of separation criterion  B (k )

Figure 3. Histogram based on gray level image
32
 24 bit true colour image of burning wood chips 256 gray level image of burning wood chips

Gray level histogram Distribution of separation criterion  B (k )

Fire Detection System 33

Reference
• A. Ollero, B. C. Arrue, J. R. Martinez and J. J. Murillo, Techniques
for Reducing False Alarms in the Infrared Forest-Fire Automatic
Detection System. Control Engineering Practice, Vol. 7, No. 1, 1999,
123-131.
• P. Illera, A. Fernandez and J.L. Casanova, Automatic Algorithm for
the Detection and Analysis of Fires by Means of NOAA AVHRR
Images, EARSeL Advances in Remote Sensing, Vol. 4, No. 3 – XII,
1995, 1-6.
• Video Smoke Detection - A New Concept, A New Technology,
http://www.vintec.com/vsd2.htm
• BS5839: Fire Detection and Alarm Systems for Buildings. Part 1:
Code of Practice for System Design, Installation and Servicing.
London, UK. 2013.
• Codes of Practice for Minimum Fire Service Installations and
Equipment and Inspection and Testing of Installations and
Equipment. Hong Kong Fire Services Department. 2012.
• A Dramatic Development in Fire Detection & Prevention,
http://www.isgfire.com/firedet2.htm

Fire Detection System 34

• J. Li, Lecture notes for fire engineering system on image processing,
2002
• N.F. Fong, MSc Lecture Notes on Advanced fire detection on image
processing electronics (2017)
• Fire Protection Engineering Handbook, 2nd Ed. National Fire
Protection Association/ Society of Fire Protection Engineers,
Quincy, MA, USA. 1995.
• B. J. Meacham, International Developments in Fire Sensor
Technology. Journal of Fire Protection Engineering, Vol. 6, No. 2,
1994, 89-98.
• M.R. Jones, M.Q. Brewstev and Y. Yamada, Application of a
Genetic Algorithm to the Optical Characterization of Propellant
Smoke, Journal of Thermophysics and Heat Transfer, Vol. 10, No. 2,
April-June, 1996, 372-377.
• G. Healey, D. Slater, T. Lin, B. Drda and D. Goedeke, A System for
Real-Time Fire Detection, IEEE, 1993, 605-606.
• E. Gorovici, What’s the Next After Digital Video Recording
Becomes the Standard? Security World, Security Information Inc.,
Korea, 1999.
• B. Aird, A. Brown, No Fire without Smoke – CCTV Breakthrough
in Fire Detection. Nuclear Engineering International, Vol. 42, No.
514, 1997, 35.
35
• J.Li, N.K. Fong, L.T. Wong, Smoke and fire image segmentation
using Shannon’s entropy method and mathematical morphology,
Shandong-Hong Kong 2003 Joint Symposium on New Technology
for Better Built Environment, 17-18 October 2003, Qingdao, China
• P. Guillemant and J. Vicente, Real-time Identification of Smoke
Images by Clustering Motions on a Fractal Curve with a Temporal
Embedding Method, Optical Engineering, Vol. 40, No. 4, April
2001, 554-563.
• D. Wieser and T. Brupbacher, Smoke Detection in Tunnels Using
Video Images, 12th International Conference on Automatic Fire
Detection, March 25-28, 2001, 1-12.
• T. Chen, Recent Developments in the Core of Digital Signal
Processing, IEEE Signal Processing Magazine, Jan. 1999, 16-31.
• F. Farbiz, M. B. Menhaj and S. A. Motamedi, Fixed Point Filter
Design for Image Enhancement Using Fuzzy Logic, IEEE, 1998,
838-842.
• F. Russo, FIRE Operators for Image Processing, Fuzzy Sets and
System, Vol. 103, No. 4, 1999, 265-275.
• J. Fan, D.K.Y. Yau, A.K. Elmagarmid and W.G. Aref, Automatic
Image Segmentation by Integrating Color-Edge Extraction and
Seeded Region Growing, IEEE, Vol. 10, No. 10, Oct. 2001, 1454-
1464.
Fire Detection System 36