Abstract:
Conventional fire detection systems use physical sensors to
detect fire. Chemical properties of particles in the air are acquired by
sensors and are used by conventional fire detection systems to raise an alarm.
However, this can also cause false alarms; for example, a person smoking in a
room may trigger a typical fire alarm system. In order to manage false alarms
of conventional fire detection systems, therefore a computer vision-based fire
detection algorithm is needed. The
algorithm can be used in parallel with conventional fire detection systems to
reduce false alarms. It can also be deployed as a stand-alone system to detect
fire by using video frames acquired through a video acquisition device. A novel
fire color model is developed in CIE L*a*b* color space to identify fire
pixels.
Flowchart:
The motivation for using CIE L*a*b* color space is because it
is perceptually uniform color space, thus making it possible to represent color
information of fire better than other color spaces.
RGB to CIE L*a*b* Color Space Conversion:
Following
equation shows conversions.
Where, Xn, Yn, and Zn are the tri-stimulus values of the
reference color white. The data range of RGB color channels is between 0 and
255 for 8-bit data representation. Meanwhile, the data ranges of L*, a*, and b*
components are [0, 100], [–110, 110], and [–110, 110], respectively.
The range of fire color can be defined as an interval of
color values between red and yellow. Since the color of fire is generally close
to red and has high illumination, we can use this property to define measures
to detect the existence of fire in an image. For a given image in CIE L*a*b*
color space, the following statistical measures for each color channel are
defined as,
Color Modeling for Fire Detection:
Where Lm*, am* and bm* are a collection of average values of
the L*, a*, and b* color channels, respectively; N is the total number of
pixels in the image; and (x, y) is spatial pixel location in an imaging grid.
The numeric color responses L*, a*, and b* are normalized to [0, 1]. It is
assumed that the fire in an image has the brightest image region and is near to
the color red. Thus, the following rules can be used to define a fire pixel:
Where, R1, R2, R3, and R4 are binary images which represent the
existence of fire in a spatial pixel location (x, y) by 1 and the non-existence
of fire by 0. R1(x, y), R2(x, y), and R3(x, y) are calculated from global
properties of the input image. R4(x, y) represents the color information of
fire.
a final fire pixel detection equation can be defined as,
Where, F(x, y) is the final decision on whether a pixel
located at spatial location (x, y) results from fire or not.
You can detect the fire using following matlab code.
Code:
% Put image fire in the same folder where this code is put.
i=imread('fire.jpg');
C = makecform('srgb2Lab');
i_Lab = applycform(i,C);
x=i_Lab(:,:,1);
y=i_Lab(:,:,2);
z=i_Lab(:,:,3);
L=sum(x);
L=sum(L');
a=sum(y);
a=sum(a');
b=sum(z);
b=sum(b');
d=zeros(150,200);
L=L/(150*200);
a=a/(150*200);
b=b/(150*200);
for p=1:1:150
for q=1:1:200
if (x(p,q)>L)
d(p,q)=1;
else d(p,q)=0;
end;
end;
end;
e=zeros(150,200);
for p=1:1:150
for q=1:1:200
if (y(p,q)>a)
e(p,q)=1;
else e(p,q)=0;
end;
end;
end;
f=zeros(150,200);
for p=1:1:150
for q=1:1:200
if (z(p,q)>b)
f(p,q)=1;
else f(p,q)=0;
end;
end;
end;
g=zeros(150,200);
for p=1:1:150
for q=1:1:200
if (y(p,q)>z(p,q))
g(p,q)=1;
else g(p,q)=0;
end;
end;
end;
h=zeros(150,200);
for p=1:1:150
for q=1:1:200
if (d(p,q)&&e(p,q)&&f(p,q)&&g(p,q)==1)
h(p,q)=1;
else h(p,q)=0;
end;
end;
end;
h=sum(h);
h=sum(h');
if (h >10 )
fprintf('FIRE DETECTED\n');
else
fprintf('FIRE NOT DETECTED\n');
end;
