Image processing techniques applied for pitting corrosion analysis

IJRET: International Journal of Research in Engineering and Technology eISSN: 2319-1163 | pISSN: 2321-7308
__________________________________________________________________________________________
Volume: 03 Issue: 01 | Jan-2014, Available @ http://www.ijret.org 385
IMAGE PROCESSING TECHNIQUES APPLIED FOR PITTING
CORROSION ANALYSIS
Tawfeeq A. Alkanhal1
1
Research Center director of Engineering and Applied Science, Faculty of Engineering, Majmaah University, Saudi
Arabia,
Abstract
In order to study the behavior of the early stage of pitting corrosion, an image analysis based on discrete wavelet packet transform
and fractals was used. Image feature parameters were extracted and analyzed to characterize the pitting corrosion development with
test time. It was found that the feature parameters: Shannon entropy, energy, fractal dimension and intercept increased with the test
time. Therefore the image processing techniques were promising and effective tools to analyze and detect the pitting corrosion.
Keywords: corrosion, pitting corrosion, surface topography, surface analysis, carbon steel, tap water
-----------------------------------------------------------------------***----------------------------------------------------------------------
1. INRRODUCTION
Pitting corrosion is localized accelerated dissolution of metals
that occurs as a result of a breakdown of the otherwise
protective passive film on the metal surface [1]. Pitting is
regarded as one of the most dangerous and an insidious form
of corrosion, since it often leads to perforation and to a
consequent premature corrosion failure [2-5]. For general
corrosion, the corrosion monitoring and the life prediction of
the metal materials can be carried out without difficulty.
However for the localized corrosion, the precise monitoring is
difficult to manage reliably. Therefore, recent research efforts
concentrate on the development of numerical descriptors and
image processing techniques to monitor the localized
corrosion.
Image analysis technique, which nowadays has become a
worthwhile tool for being able to perform analyzes of fast,
inexpensive and non-destructive for many processes. Image
analysis is an appropriate tool to characterize qualitatively and
quantitatively the early stage of the mechanism of damage by
corrosion [6-9]. Pitting corrosion through pit numbers
determination and its morphology characteristics was
evaluated using image analysis [10-13]. The use of image
analysis of corroded surfaces to determine the morphology
and extent of pitting corrosion has not received much
attention.
The objective of the present work is to use fractal and wavelet
analysis to assess the growth of pitting corrosion with the test
time for low carbon alloy steel.
2. IMAGE ANALYSIS TECHNIQUES
Various methods exist for characterization of surface
topography and extract the relevant features. Wavelet
transforms and fractals were adopted in this study, since,
unlike other methods, they characterize surface topographical
features over different scales. They are described briefly
below.
2.1 Wavelet Packet Decomposition
The main advantage of using wavelets is that it provides
multi-resolution analysis. Multi-resolution processing can
improve the image quality obtained from microscopy
techniques, such as SEM and others. Wavelet decomposition
and its extension, wavelet packet decomposition, have gained
popular applications in the field of signal/image processing
and classification because of many outstanding properties of
wavelet packet. Wavelet transforms enable the decomposition
of the image into different frequency subbands, similar to the
way the human visual system operates [14].
In 2-D discrete wavelet packet transforms (2-D DWPT); an
image is split into one approximation and three detail images.
The approximation are then split into a second-level
approximation and detail images, and the process is
recursively repeated. The standard 2-D DWPT can be
described by a pair of filters; a low-pass filter h and a high-
pass filter g [15]. The 2D discrete wavelet packet
decomposition of an M x N discrete image x up to level p+1
))logmin(log0( 22 MNP  is recursively defined in
terms of the coefficients at level p as follows[14]:
P
jnimKm n
P
jiK cnhmhc )2,2(,
1
),(,4 )()( 

 
(1)
P
jnimKm n
P
jiK cngmhc )2,2(,
1
),(,14 )()( 

  
(2)

__________________________________________________________________________________________
P
jnimKm n
P
jiK cnhmgc )2,2(,
1
),(,24 )()( 

  
(3)
P
jnimKm n
P
jiK cngmgc )2,2(,
1
),(,34 )()( 

  
(4)
Where
0
0c is image x and K is an index of the nodes in the
wavelet packet tree denoting each subband; h and g are the
filter coefficients of low-pass and high-pass filters,
respectively. Supposing that Haar basis has been used, h={-
0:7071; 0:7071}, and g = {0:7071; 0:7071}. At each step, the
image
P
Kc is decomposed into four quarter-size images
1
34
1
24
1
14
1
4 ,,, 





 P
K
P
K
P
K
P
K cccc . The capital letters (N or M) are
maximum constants defined by the image size. However,
small letters (m or n) are defined at each step. For example,
when P = 5, P can be 1, 2, 3, or 4 and so on.
The Shannon entropy and the energy in different sub-bands
are computed from the sub-band coefficient matrix as:
2
),(,
2
),(, log)( P
jiki j
P
jikP cckEntropy 
(5)
2
),(,)(  i j
P
jikP ckEnergy
(6)
Where Energy p(k) and Entropy p(k) are the energy and
entropy of the image projected onto the subspace at node (p,
k). The entropy of each sub-band provides a measure of the
image characteristics in that sub-band. The energy
distribution has important discriminatory properties for images
and as such can be used as a feature for texture classification.
From the equations above, it follows that the wavelet entropy
is minimum when the image represents an ordered activity
characterized by a narrow frequency distribution, whereas the
entropy is high when an image contains a broad spectrum of
frequency distribution.
2.2. Fractal Analysis
Fractal geometry is a well-known non-traditional method
which has found many applications in science and
engineering. It is common knowledge that many objects in
nature are of irregular form which cannot be described by
Euclidean geometry. These non-Euclidean objects are called
fractals, and can be described by non-integer numbers. These
non-integer numbers define the fractal dimension (FD) of an
object. The main concept of fractal geometry analysis is that a
fractal dimension can be considered as a quantitative measure
of object surface heterogeneity because of its inherent self-
similarity features. In a simplified representation, one could
interpret the fractal dimension as a measure of heterogeneity
of a set of points on a plane, or in space. FD can numerically
characterize the variation in surface structure caused by
corrosion [11], which corresponds to morphology changes in
grey value images captured by microscopy techniques such as
SEM or AFM.
A number of methods have been used to calculate fractal
features of the surfaces such as Fourier, Kolmogorov, Korcak,
Minkowski, root mean square, Slit Island, etc. These
techniques differ in computational efficiency, numerical
precision and estimation boundary. The most efficient
procedure for measurement of the FD of surfaces, and one
which allows characterization of anisotropic surface as well,
seems to be through Fourier analysis [e.g. 16, 17]. Therefore,
Fourier analysis is adopted to estimate fractal values in this
work. For a surface image represented by the function I(x, y),
the power spectral density PSD is equal to the square of the
Fourier transformation F(u, v) of the surface function I(x, y).
The power spectral density function is defined as;
2
),(),( vuFvuS 
(7)
Where u and v are the spatial frequencies (number of waves
per unit wave length) in the x and y directions respectively.
The PSD is converted to the polar coordinate system S (f),
such that
22
vuf  . The value of S(f), at each radial
frequency f, is averaged over angular distributions. The slope
of the linear regression line  is related to FD by equation
[18]:
2
8 
FD
(8)
It is reported in the related literature that fractal dimension and
intercept are significant fractal parameters that describe the
irregularity and complexity of the surfaces. Moreover, the
intercept correlates well with the overall magnitude of
roughness of the observed texture appearance of the surface
images. In the present study, fractal analysis is going to be
used to assign numerical values to indicate the development of
pitting corrosion.
3. EXPERIMENTAL PROCEDURE
Carbon steel is the most widely used engineering material, so
the cost of dealing with corrosion of carbon steels is a
significant portion of the total cost of corrosion. Carbon steel
test coupons with dimensions of 50×30×1 mm were cut from a
single sheet to ensure metallurgical uniformity. The surface
roughness is known to play an important role in developing
the corrosion [28,29], Therefore the coupon's working faces
were polished with grade 800 silicon carbide paper. A typical

__________________________________________________________________________________________
polished surface is shown in Fig.1. The inclined vertical lines
shown in the photo are the traces of polishing
A rotating corrosion-test equipment was used for studying
pitting corrosion behavior. The coupons were immersed in
1200 ml open pot having 700 ml of tap water. The pot was
fitted with four vertical baffles to break up the rotational flow
pattern. The pot and the baffles were made from plastic
material. Each coupon was attached to a rotating shaft with
pvc washers and a plastic screw to avoid the galvanic
corrosion. Because of the rotational speed, turbulence was
generated in the test water, together with a tendency for a rise
in test water temperature. Since the test water temperature
markedly affects the corrosion rate [27], the test water was
changed every one hour. The temperatures at the beginning
and the end of test were 23 oC and 24 oC, respectively. Table
1 shows the composition of test water determined by chemical
analysis.
Coupons were rotated at 160 rpm (linear velocity of 0.5 m/s)
for test periods of 1, 3, 5, and 7 h. At the end of test period the
rotating shaft with coupon was withdrawn, and air dried. The
corroded surfaces were examined by a Scanning electron
microscope (SEM). These digitized SEM images were used in
the image analysis to develop features to characterize the pits
development with the time.
Table 1 Chemical analysis of test water
Elements Na+
Ca2+
Mg2+
K+
HCO3
−
Cl−
Fl−
NO3
−
SO4
−2
pH TDS
37.8 75.2 6.8 2.7 70.2 73.3 0.28 2.6 10.7 7.2 350
Note: all values except pH are in mg/l
Fig.1: Showing the polished surface before test
4. RESULTS AND DISCUSSION
4.1 Morphology of Pitting Corrosion
The typical corrosion pits formed on the polished surfaces
rotated in tap water at different times are shown in Fig. 2. It
can be observed that many pits are formed on the surface and
each pit formed has a track accompanied it which appears as
black in photos and which looks like a comet streak tail. The
characteristic features of these pits are the presence of two
parts: a cavity at the pit centre and a rough circular band
around the cavity, which is labeled 1, as well as circular rings,
labeled 2. The same shape for these pits has been observed

__________________________________________________________________________________________
before in the literature on corrosion tests [19-23]. The comet
tails were formed in opposite direction to the rotation of the
specimen. The comet tails take the form of cone tubular
structure. The circular base of cone is the ring that formed
around the inclusion. The effect of time on the pitting
corrosion development for rotation test is illustrated from the
photo, where the pit and the comet size increase with the time.
It can be seen in these photos that many pits formed and
material removed along the track direction. This gives
evidence that the corrosion products are aggressive. This is in
agreement with that reported in Ref. [23].
Fig 2 Continue

__________________________________________________________________________________________
Fig 2 Continue
4.2 Feature Extraction of Pits Development
In the present section, the features parameters of images such
as wavelet entropy, wavelet energy, and fractal dimension and
intercept value are extracted. The results of these feature
parameters, performed by using 2D-DWPT and fractal
analysis and described in Ch.2 are given in Figs 3-6. Shannon
entropies and energy of wavelet packet decomposition of
image were calculated with the testing time from all the
selected subbands of the image as shown in Figs. 3 and 4.
Figures 5 and 6 show the fractal dimensions and intercept for
the image of corroded surface at different times. From these
figures, it is clear that the value of the entropy, energy, fractal
dimension and intercept increases with time. Each feature
corresponds to a visually recognizable property of the image
described in Sec. 4.1. That is, for undamaged surfaces which
have homogeneous textures have relatively low values of
entropy, energy, fractal dimension and intercept. While, these
extracted features have high values for corroded surfaces due
to the coarseness of the corroded surface textures.
Fig. 3 Change of Shannon entropy with test time
0 1 2 3 4 5 6 7
1.4
1.6
1.8
2
2.2
2.4
2.6
2.8
x 10
10
Entropy
Time, h

__________________________________________________________________________________________
Fig. 4 Change of energy with test time
Fig. 5 change of fractal dimension with test time
Fig. 6 change of fractal intercept with test time
0 1 2 3 4 5 6 7
1.95
2
2.05
2.1
2.15
2.2
x 10
6
Energy
Time, h
0 1 2 3 4 5 6 7
1
1.5
2
2.5
FractalDimenssion
Time, h
0 1 2 3 4 5 6 7
22
23
24
25
26
27
28
Intercept
Time, h

__________________________________________________________________________________________
CONCLUSIONS
Image processing based on discrete wavelet transform and
fractal analysis was used to characterize the corroded damage
images. The following results can be drawn:
(1) The extracted feature parameters; Shannon entropy,
energy loss, fractal dimension and fractal intercept
increase with exposure time..
(2) The results indicate that the image analysis
procedures are promising techniques since they are
effective in characterizing the changes of surface
topography with exposure time.
(3) The surface topography with test time showed that a
tubular corrosion product structure developed with
corrosion pit in reverse direction of rotation.
REFERENCES
[1] Frankel, G. S.: Pitting Corrosion of Metals A Review of
the Critical Factors. J. Electroch. Soc. 145(6), pp. 2186-
2198, (1998)
[2] Davis, R.J.: Corrosion: Understanding the Basics. ASM
International, (2000)
[3] Bardal, E.: Corrosion and Protection. Springler verlag,
(2004)
[4] Jones, D. : Principles and Prevention of Corrosion",
Prentice-Hall, (1996)
[5] MAlik,A.U., Al Fawzan, S.: Localized corrosion Of
AISI 316L SS in Arabian Gulf seawater, Desalination,
123, pp. 205-213, (1999)
[6] Li,G.B., Li, T.J.: Extracting characteristics from
corrosion surface of carbon steel based on WPT and
SVD,Z. Qian et al. (Eds.): Recent Advances in CSIE
2011, LNEE, Springer-Verlag Berlin Heidelberg, pp.
105-112, (2012)
[7] Gao,Z.M, Han, X. B., Dang, L.H., Wang, Y., Bi, H.C.:
Evaluation of simulated atmospheric corrosion of Q235
steel by wavelet packet image analysis, Int. J.
Electrochem. Sci., 7, pp. 9202-9212, (2012)
[8] Tao,L. , Song,S.Z., Zhang,X.Y. , Zhang, Z., Lu
,F.:Image analysis of atmospheric corrosion of field
exposure high strength aluminum alloys, Appl. Surf.
Sci., 254, pp. 6870-6874,(2008)
[9] Wang, S., Song,S.: Image analysis of atmospheric
corrosion exposure of zinc, Mater. Sci. Eng. A 385, pp.
377–381, (2004)
[10] Pereira, M. C., Silva,J.W. J. , Acciari,H.A., Codaro,E.
N. , Hein L.R. O. : Morphology Characterization and
Kinetics Evaluation of Pitting Corrosion of
Commercially Pure Aluminium by Digital Image
Analysis, Materials Sciences and Applications, 3,pp.
287-293, (2012
[11] Pidaparti, R.M., Aghazadeh, B.S., Whitfield ,A., Rao,
A.S., Mercier, G.P. : Classification of corrosion defects
in NiAl bronze through image analysis. Corrosion
Science, 52, pp.3661–3666, (2010).
[12] Chenghao, Li., Wei, ZA. : Fractal characteristic of pit
distribution on 304 stainless steel corroded surface and
its application in corrosion diagnosis, J. Wuhan Univ.
Techn-mater, Sci. Ed., pp.389-393, 920070
[13] Codaro, E.N., Nakazato, R.Z., Horovistiz, A.L., Ribeiro
L.M.F., Ribeiro,R.B., Hein, L.R.O.: An image
processing method for morphology characterization and
pitting corrosion evaluation, Mater. Sci. Eng. A 334,
pp. 298-306, (2002)
[14] Huang, K., Aviyente, S.: Wavelet Feature Selection for
Image Classification IEEE Trans. Image Process.,
17(9), pp. 1709-1720, (2008)
[15] Mallat, S.:A wavelet tour of signal processing,
Academic, New York, (1999)
[16] Russ, J. C. :Fractal analysis, Encyclopedia of Material:
Science of Technology, Elsevier Science Ltd., 3247–
3254(2001)
[17] Zhang, J.: Detection and monitoring of wear using
imaging methods, Ph.D. thesis, University of Twente,
Enschede, The Netherlands, (2006)
[18] Babadagli, T., Develi, K.: Fractal analysis of natural
and synthetic fracture surfaces of geothermal reservoir
rocks, Proc. World Geothermal Congress 2000
Kyushu - Tohoku, Japan, May 28 - June 10 ( 2000)
[19] Budiansky, N.D., Hudson, J.L., Scully, J.R.: Origins of
persistent interactions among localized corrosion sites,
Symp. in Honor of Hans Bohni, Electrochemical
Society, Salt Lake City, UT, (2002)
[20] Wang, W., Zhang, X., Wang, J.: Pits with colored halos
formed on 1Cr18Ni9Ti stainless steel surface after
ennoblement in seawater. Mater. Sci. Eng. C 29, pp.
851–855, (2009)
[21] Karrab,S.A., Doheim, M. A., Mohamed S. Mohammed,
Ahmed,S.M.: Study of cavitation erosion pits on 1045
carbon steel surface in corrosive waters. ASME, J.
Tribol 134, 0011602- pp. 1-6, (2012)
[22] Karrab,S.A., Doheim, M. A., Mohamed S. Mohammed,
Ahmed,S.M.: Investigation of the ring area formed
around cavitation erosion pits on the surface of carbon
steel. Tribol Lett 45, pp. 437-444, ( 2012)
[23] Alkanhal Tawfeeq A. , Osman M., Ahmed, S. M.:
investigation into tubular structure formed by pitting
corrosion on the surface of carbon steel, J. Eng. Sci.,
(Assiut Univ, Egypt) 33 (6), 2165 (2005)

Image processing techniques applied for pitting corrosion analysis

More Related Content

What's hot (18)

Viewers also liked (17)

Similar to Image processing techniques applied for pitting corrosion analysis (20)

More from eSAT Journals (20)

Recently uploaded (20)

Image processing techniques applied for pitting corrosion analysis