You are on page 1of 4

The 16th International Conference on Electrical Engineering/Electronics, Computer, Telecommunications and Information Technology (ECTI-CON 2019)


  
     

       

 
  !
 ' -

.
+ ) / 0 

+
#
Department of Electrical Engineering Department of Electrical Engineering Department of Electrical Engineering
Faculty of Engineering Prince of Faculty of Engineering, Prince of Faculty of Engineering, Prince of
Songkla University Songkla University Songkla University
"  #
$ 

 %&' "  #
$ 

 %&' "  #
$ 

 %&'
(&&'&&)'*+,
, 

,*+,
, / 0,1*+,
,

Abstract—Nowadays, flooding is one kind of hazard that 8(9$       
   
<  #
usually occurred all over the world, which brings enormous / 

 

   
  
 
$
damage to human life and property. In order to decrease the 0 
   #  /    1    ,
harm of flooding, the image edge detection techniques have been 5     +
  $  4
    
used to monitoring the state of flood area in the real-time in this   0 # 1 
     #  1$
paper. The processing time of the method is a key point that ;0$
 " 1,   8?9 +    
should be considered, which related to the timeliness of the flood    

 +

   $ 0 
 
monitoring system. And the accuracy of the predicted flood area  

  1   
   

is also the standard that can be used for evaluation. The

   
 
 0   
  ,
experimental results have been affected by the changing of the
parameters of each algorithm and the method that has the
0         $  + 
highest accuracy and lowest processing time have been expected    4
   1 1
  


one. According to the results, Canny edge detection has the best     ,
performance, which has been considered for implementation on   +
+$    0
++
  
 

the real hardware in the future for the outdoor environment.  + 
$         
+        

, 5 

 <
Keywords—image edge detection techniques, flood monitoring

  +
         $ 
system, Canny edge detection, Roberts edge detection, Prewitt edge
detection, Laplacian of Gaussian edge detection
 +   
 0 # 1
    0
+  
  

 ,    
 1 
, -2342-   
 
0
         

    

     1  0,
5 0
 # 0$     

 
 0 

1 
 1  

#$ 
  
+ $  
   +
+     0 +
, "  
, 6 0$   
 

   
 @+  
        

 

 1


 
   $ 0 
#  
  

 #


 +  
+ +$   1 
 1  

, 7 0 
 1   

  + 
  

     
         , "   +       
+ + 

,     5A5 
 
    0

  + 
 0 1 ,   $ "  E

   
 1      
      
 
 ,
8)9,        
    + 
$ 0 
  1 +  
    , , 622A2BF
5     $   
   
1 
 
 0 # 1
  

             

0 


 ,  8:9$  
 +   
 1
++
      
 0, 4 $ 
+         
,    

 
 #       0 


 
+    ; 

,   
  +  1        
++
    
 


  0,     +
     $  +  , 6 0$  4

  A B    
4
     
   


  0  0
1   +
     
  
,
+   
      0
 - 
<   $  4
  $ A B  
   0
 $ 0   
1  +   + +
      G  1
 ;0


0
       ,   

    1 
  0  
 +


 $    4
 $ 
 
   $  ,     

 

++
 ,
=
  
     
  

0 A. Image Edge Detection Techniques


   8>98?9,   8>9     
        + 
   
     +   
 +   + 
 
 +  $ 0 
     
   #  
   # 
#$ 111
 
1/   
 
  ,   + $ 
, - 
 $ 7   
  0  
 1  
  +@   
 
 1 



    1   1
  
 

 
 +@ 
 1  +
 0   1  ,  

+
, =        
 1
++

   10  +@
   1  
 @
  
 1   
    

 
 $    
 1   +

  +@

 10  
  
 
$ 0 


   +
 8>9,
   , 5    $  A
+

 B


CA BD       
    + 
,  Canny edge detection:

%?H?'H))(:I%IJ),&& K'&%  %'?

Authorized licensed use limited to: Univ of Calif Santa Barbara. Downloaded on June 28,2020 at 11:41:58 UTC from IEEE Xplore. Restrictions apply.
The 16th International Conference on Electrical Engineering/Electronics, Computer, Telecommunications and Information Technology (ECTI-CON 2019)

4
     
     $ 0  ;0        0 

      
   
 
 0   
,   < 

 
  $ 0 

1

 
 
 +
 1
 
     +
 ,       8%9,  
 0 ) @ ) #  0

 +   +   0   , 8H9,  , :,   #  +   #  
 1 90° ,

   B
  +     & 

& & &  & 


 1  
  - 
@
++ 
    & 
 , , 4
    
  + 
 
 '   & 
 , :, )@) ] ;0
& & & ' & '   , :$     +@  
 1
 +
  0  
    1
 
 '   &       #
 <
  

      
 
 
   1  
,
  
++ @
 
 

 1 1
   C'D,
 , ', )@) ] 4
 Laplacian of Gaussian Method:
  1  $  
 
 1    
  +    G 4
 $  1$
 ;0
     1  B

 ,  0 

  #          
$ 0
 
 B

     CD, (, ) 
    A
+

   0 
  <    1
 
0   +@   +   (, ), ^ 
+

 
    
, 5  A
+

   
  <<    
,     +$ 
1    $   B

 1 
++
 
 
 
 1  + 1  ' 
 1   + A
+

  
      8(9,
( ,  )  
  ,  
  


 1 

   C'D 0    
 1   
C)D 8(9, $   
@
++      
  +     ,   $   1  

   +, 


 0  G 

',   
 
 ', =  
  +@   C
D C1D CD
     
 '$  +@ 
 1 #+
 1 
 
_  `   +@, =  
  +@  10  , >, 


   A
+


 0  $ 0 
 1 
  _0
#` , 
 + 
$  0
# +@ 
 1 #+  0      A
+

   0   , >, CD  
      ,  
   
 
$ 0  , > C
D  0
   +
 0 
 
 
1+  CD,  bCD 
 
 ( )  , > C1D    
 CD$ 0  0 
(, ) =
 CD

@     
  +   
1+
|| = | | +   C'D 
        , > C
D, = bbCD    

 CD$      , > CD +   <

 = arctan ( ) C)D   $ 0 
    +
    
 
,
  
A
+

   0  C>D
  + 1
Roberts Method: # A
+

   0   , (,
   
 1     
 +

  
 Laplacian(f) = + C>D
 
#$ 0   0   , ) 8%9,  
 

 
 1     +

 
  &  &
+@   
 0  +  #, 

++ @
 
 
 1    C'D$ 1   : 

     C:D$ 0 
  45°
     


     +@, &  &

 & &   , (,  + 1 # A


+



&   & B. System Analysis


         +
+
 4
 $
   1$ ;0$
 A B$ 0
 0    
 , ), '@' ]  1

 +  ,    

  + 


  $ 


  +   
 1
 
 = arctan     C:D 
  

   +
 ,   +
 
   
 1 

   /   
1
Prewitt Method:

%'H

Authorized licensed use limited to: Univ of Calif Santa Barbara. Downloaded on June 28,2020 at 11:41:58 UTC from IEEE Xplore. Restrictions apply.
The 16th International Conference on Electrical Engineering/Electronics, Computer, Telecommunications and Information Technology (ECTI-CON 2019)

          


 1    
   

 
 
  

 , +          0 #, 5  

$         +
  $    
     
@     
 
       
 1
   
 
,  @

   +


  
 $ 0 
 +      0   + 


1 

  +  
 

  
 
  0    1
   @ + 
G  

 10  +


 
 
  -   - (,  
   1
#
   + 
8:98>98&9,    $  + 
 

 +     
  0

 


++          +




  

  ,
+ 
 
 $ 0     
 
 1


   
 

 
 
 , 
1
  0        
@ 
 
++  

,   +  C;D   
  +  


    + 

   
 +   0  C
D C1D CD CD


  

      
, "
 $   , ?, 5 @+ +  @
+ 0  4
   0
   &,'   
  0


 C-D 
      

  ,  

 +  C;D$    

    $  +   @+   0   , ?$ 0 
0  +    

    

 @
+  4
       0 

 C-D,    
 
      
  
 0     &,',  , ? C
D  
+@
 

 +  + +       
 + 
, 5  ++  #  
1    
 +@ 1$ 0   0  C(D,  B 


 
$  4
      
 

   +
, 
++   , ?C1D, $   +      
     10     +

  

5A , 6 4A5""452- "F" 2 5-5AF!-B 6  
  1 #,    1   
$ 
544354F

  1 # 
 1 
   

  , ? CD,
Predicted Pixel  , ?CD     
0    
 

Confusion Matrix  
 

  
  
 
 1  +
 0 
Not Flood Area Flood Area
+ 
   , ?CD   +

 ,  
5
 -    5
- ;
4
    $   1$ ;0
 A B  
;@   5
- ; 

 1 +  0  + ,
!"
Accuracy = C(D
"!#!#"!

 +   


  

  1
  
  
  + 
 0 0 $ 0 
 
  +   
   

  @+        
1
   
 
 1 +   
  
    ,   @+$  + 
             

     5A5    +,
 , H, 5
 CkD 8
 +   CD9 0  4
  
, j;-5A "3A"   

1       
  @+$    
  +

1



+
1        $  @ +


 
+     ; 

  1
  0   
#
   0 
    0
 # 
 $ 
 
  , 


   
 
    
$ 0
1  
 
 ,    
$  
 0
  
   0 
    $ 0
 
 0 
++
     
  
, =
 
   $  ++   
  0

1     ,   
$        , %, 5
 CkD 8
 +   CD9 0   1  
  
,  $  + 
       

1       
 0
    
    
 +,  <

  (&&@&': +@ 
+    1  , H +  



++  4

 
  +    @+  +     , 7  0 
   

   
CD 4  CD >H'>&3 4;3 * ,(&B6<$ 5 HB, 
1$   0

1   # &,'
 &,)
 
 
  
,   , %$   

 
 , H
  , %
  @+   0
    1  $   
 
 
 
         G 4

  1,   &,',      0   
 

1 +

   +
+    $ 0  &,$ 0   
     ;0  

 

    0  
 ,  
   ,   A B  $    
  
       &  ,   +  

 
  0   

     


  
 
1       

 
 >&k, =   
@    +  

%'%

Authorized licensed use limited to: Univ of Calif Santa Barbara. Downloaded on June 28,2020 at 11:41:58 UTC from IEEE Xplore. Restrictions apply.
The 16th International Conference on Electrical Engineering/Electronics, Computer, Telecommunications and Information Technology (ECTI-CON 2019)

  1$ ;0$


 A B      
 ('k$ 0
 E, 42-4A3"2-
    

   4
     5   
       
 1

 ?>k,

 <
    5A5    + 



 0  

1  $  4

     1   0 
  1 + 

0   


  0 +  $ 0
-  - ' - )  

  + 1  1 +   
0

 
      0 #,
54]-2=AB-
- : - > - (  
 0
 ++  1 
  
+
  
   $ ; "  #

 , &, 2 
 + 
 
 1 
+    0

  
 +

3 , 
  0  #  
#
   

   +
 0
  
,
 @  
 + 

  0   ,&$ 
0  
    -   - >
 
+    -4"

  $ 0  
 - (   
  
#, - 89 ;, 
 $ , 6 
$ B, "
$ A, ; $ A, 6 
$
 6,
 +  
 
  
 7 1  
 B, "
/$ _ 0


 
 "51
     

 
 ++
++
   0
 
, =  G A  
   0 
 $` ; 

4  
$ ;
 5II4$  , )($  , ?qH$ ++, ':q'>'$
- '
 (  0   0
 
  
 0 1  '&,
 
 
    ,    
 - )$   8'9 , ; +$ A, $
 , 4


$ _ 

   1


+   
 
 0

  
++
,  35E 
  $`  "    $ 4  

4 + C4"44D$ '&> % 
 
 4  $ ++, ?>)q
5A , "3A" 2 6 6B6" 544354F 5- ?>H$ - 1 '&>,
42";2--B ;24""-B  2 546 46-3 5- 8)9 5,  # $ , 4, 6v<$ , " $
 ],6, . $ _
  
- -E2--     
$`  
   "  $
Edge Detection Methods 42- '&>: 5
 4    $ ++, &&:&H'q
Accuracy (%) &&:&H>$ - 1 '&>,
[Processing Time (s)] Canny Roberts Prewitt LoG
8:9 !
 $ .
+ -$ 
    $
 

+
# $
97.40 (,%: >(,&& )H,>H _
  
 ;  1
 
      $
-  ` 
 
 4    "
 
 $ 

[2.15] 81.879 81.919 8,%:9

 5++
 , '&H,
96.05 >',( >',) )H,: 8>9 w, x<z#
 , 5#$ _4 +
     
 
 @

   
 +  $` ; 
" 


- ' [2.12] 81.859 82.259 8',H&9 
 
 "$  , %>$ ++, '(?>q'(H'$ '&>,
:H,% 8(9 -, 
 $ 5, "
$ 5, =

$ 5, 511
#
$ ", ]

$
 ", ,
98.77 :H,% :H,%
!
$ _
  
   
    G

- ) [1.95] 81.859 82.259 8',H%9
++
   # / 

 

    $` . 

 

 5++  
    $  , ?$  , $ ++,
90.64 :,?H :&,%> '?,& H?q%($ '&>,
- : [2.31] 82.429 82.349 8',&H9 8?9 ", 5$ , "
#

$
 4, 3, ]
$ _    5 


++
    B "+

   $`  '&? 
96.72 :(,'? :(,): ::,:' 
 
 4    4 +
 
  

- > [2.19] 82.019 82.069 8',)?9 4 + 
 C444D$ ++, q:$ 1 '&?,
8H9 F, !
$ B, =
 $ 6, F$
 B, =$ _
  
++
 
76.15 )%,)> )%,: )H,>&          35E    
- ( 82.899 82.449 
  $`  
 
 5 
  C45D$ '&?
[2.21] 8',)&9
 
 
 4  $ ++, &q:$ 5  '&?,
8%9 , "

 ", ;$ _ 
 
  
     
   
1 '
   


    

1 "#$`  " 
 ;  $
 +  +   
    
 
 4 
 $ ; 0
 1 "  C"42;"D$ '&(
0
   , 7 
 
  4
   
 
 4  $ ++, '':q''H$ 2 1 '&(,
    
   

 
 
 , 8&9 j,  $ , A$ F,  $
 ", 


$ _5 +   
 +   
      

    1
 + 1
1
  1
       
@
 
 
  + 1$`  
  "$  


    
 )  $ 0 
 1 

  
 4 + "   "  5++
 $  , ):&$ ++, '>&q

++
  1
   +     '($ 
'&(,

 ,

%)&

Authorized licensed use limited to: Univ of Calif Santa Barbara. Downloaded on June 28,2020 at 11:41:58 UTC from IEEE Xplore. Restrictions apply.

You might also like