The Ninth International Conference on Electronic Measurement & Instruments ICEMI’2009

Study of Image-Based Pavement Cracking Measurement techniques

Sun Zhaoyun1, Wang Chaofan2, Sha Aimin2
1
College of Information Engineering, Chang’an University, Xi’an, China, 710064
2
Key Laboratory for Special Area Highway Engineering of Ministry of Education, Chang’an University, Xi’an, China,
710064
E-mail :w_ruofan@163.com

Abstract –Cracking is one of the most common diseases that entirely for most cracks in actual pavement surface
happen to the pavement and a field cracking survey and a images are disconnected and unclear.[6]
forensic investigation are important work for preventive Some of the image characteristics including vertical
maintenance. In order to measure the pavement crackings histogram, horizontal histogram are calculated to classify
automatically, this paper uses digital image processing
the longitudinal, transverse, block and alligator cracks.[7]
techniques to process the pavement cracking images acquired
by the digital camera and classify the pavement cracking types
This paper uses digital image processing techniques to
according to the topological properties to line crackings and process the pavement cracking images acquired by the
alligator cracking, then classify the line crackings to digital camera and classify the pavement crackling types
longitudinal and transverse cracking according to the chain according to the topological properties to line cracking
code of the cracking in the image. At the same time, the paper and alligator cracking, and then classify the line cracking
calculates the length and width of the line crackings and the to longitudinal and transverse cracking according to the
area of the box rectangle of the alligator cracking. Finally, the chain code of the cracking in the image. At the same time,
paper calculates the Cracking Index (CI) of the pavement. the paper calculates the length and width of the line
Keywords –image processing, chain code, pavement cracking, cracking and the area of the box rectangle of the alligator
preventive maintenance, image segmentation
cracking. Finally, the paper calculates the Cracking
Index (CI) of the pavement.
I. INTRODUCTION
II. PAVEMENT CRACKING IMAGE PROCESSING
Cracking is one of the most common diseases that
affect the early performance of the pavement. And the In order to get good images for further processing, the
conventional pavement cracking detection methods used pavement cracking images are acquired by a digital
in present specification are manual, which are time camera in good lighting condition. One of the most
consuming, dangerous, labor intensive and subjective. common defects of cracking images is poor contrast
Along with the development of the digital image resulting from a reduced image amplitude range and this
processing techniques, the automatic monitoring and can be improved by contrast manipulation, which is
nondestructive testing methods based on image amplitude rescaling of each pixel. The luminance
processing are used more and more frequently. histogram of a typical cracking image that has been
Many types of equipment are developed for rapid
linearly quantized is usually highly skewed toward the
monitoring and evaluating of road pavements, such as
darker levels and one means of enhancing these types of
AMAC reg[1], Swedish PAVUE, CSIRO crack detection
system, American ARIATM system, Canadian ARAN
images is the histogram modification technique, in which
system, Japanese KOMATSU system, and so on. They are the original image is rescaled so that the histogram of the
very powerful tools to obtain the surface cracking map. enhanced image follows some desired form. The
Besides, many researchers have studied the recognition and cracking images may be subject to noise and interference,
classification of pavement cracking images and imaging mainly the photographic grain noise, and these noise
method is widely used to identify pavement surface effects can be reduced by median filtering techniques.
distresses. The mask filtrating method are used to enhance After preprocessing of the cracking images, the main
the pavement cracking images[2] and the phase grouping operations to the cracking images are cracking detection
method[3], the modified median-filter algorithm with four and binary morphological processing.
structural elements and morphological operators[4], fractal One of the cracking images acquired is shown in
theory[5] and the maximum of class distance method[2] are figure 1.
used to extract the cracking edges and fill the gaps of the
cracking. Some novel cracking automatic detection A. Cracking detection
approach based on segment extending for complex
pavement images is proposed to resolve the problem that Local discontinuities in image luminance from one
image processing methods on pixel or connected domain level to another are called luminance edges. The
level could not identify pavement cracking correctly and crackings are changes in the pavement image amplitude
_____________________________ attribute of luminance value, which are fundamentally
978-1-4244-3864-8/09/$25.00 ©2009 IEEE important characteristics of the pavement image. The

2-140
The Ninth International Conference on Electronic Measurement & Instruments ICEMI’2009
one-dimensional profile of a cracking is shown in figure The skeleton of the cracking can be used to describe
2. its structure. Thinned crackings sometimes have the
appearance of a skeleton, but they are not always
uniquely defined. With the morphological skeletonizing
operation, the skeleton of the cracking image can be
obtained.

Fig. 1. Cracking image

Fig. 2. Line models of one-dimensional, continuous domain edge Fig. 4. Binary image of the cracking after morphological filtering

The pavement image contains the white lines besides

the crackings and the white lines should be removed by C. Gap filling operation
the pixel manipulation. If the pixel value of the pavement
image is in a certain scale, then the pixel values are Sometimes there exist some gaps between the
marked with a fixed value. This paper uses a conditional crackings. To automatically link the extremities of the
thresholding method to detection the crackings. In this crackings together depends on their relative positions and
method, the thresholding value is calculated without the orientations. This papers studies the distance and the
pixels marked. angle constraints to fill the gap between the crackings.
The fill gap distance must be specified in pixel, and the
B. Binary image morphological processing distance values should not exceed 5 pixels. However, if
necessary, you can also specify distance according to
After the thresholding operation to the cracking your need. The angle to fill the cracking gap is [-10°, 10°].
images, the binary cracking images are obtained. Also, The constraints and the gap filling course are shown in
the binary images including some isolated noise points. figure 5. The cracking images after the gap filling
Binary image morphological operations are based on the operation is shown in figure 6.
geometrical relationship or connectivity of pixels that are
deemed to be of the same class. Dilation, erosion and
skeletonization are three fundamental morphological
operations. This paper uses the morphological filtering
operations to remove most of these noise points and the
morphological filtering is combination of dilation and
erosion operations. The binary image of figure 1 is
shown in figure 3 and the binary image after
morphological filtering is shown in figure 4.
Fig. 5. Cracking gap filling

Fig. 6. Cracking image after gap filling operation

Fig. 3. Binary image of the cracking

2-141
The Ninth International Conference on Electronic Measurement & Instruments ICEMI’2009
III. PAVEMENT CRACKING TYPE
CLASSIFICATION

This paper first classifies the pavement cracksing to

line cracking and alligator cracking according to the
topological properties, and then classify the line
crackings to longitudinal and transverse cracking
according to the chain code of the cracking in the image. Fig. 8. Alligator cracking

A. Topological properties and chain code of pavement

cracking

Topological properties of pavement crackings are

invariant under rubber sheet transformation. Such a
transformation or mapping can be visualized as the
stretching of a rubber sheet containing the image of an
Fig. 9. Longitudinal cracking
object of a given shape to produce some spatially
distorted object. Connectivity is a topological properties
and connected components of an object may contain
holes. Connectivity and holes are the two main
topological properties of the crackings.
The chain code used in this paper is shown in figure 7
and the ‘*’ is the current pixel. This paper uses the chain
code to classify the line crackings and calculate the
length of the line crackings. Fig. 10. Transverse cracking

IV. PAVEMENT CRACKING FEATURE

CALCULATION

The features of the line crackings are different with

that of alligator crackings. So the features of the cracking
Fig. 7. The sketch of the chain code are calculate separately.

A. Line cracking
B. Pavement cracking classification
The features of line crackings mainly are the length
The pavement crackings are classified to line cracking and the equivalent width. The length of the cracking is
and alligator cracking according to the topological calculated through the chain code. The length of a
properties of the crackings. If the number of the holes longitudinal cracking in pixel is expressed as IL and the
that the cracking contains is no less than 1and the number length of a transverse cracking is IT.
of the connectivity is no less than 1, then the type of the
cracking is alligator cracking; if the number of holes is 0 ,
and the number of the connectivity is no less than 1, and In which N0 is the number of the zeros in the chain
the type of the cracking is line cracking. code of a cracking, N1 is the number of the ones in the
The line crackings are classified to longitudinal and chain code of a cracking and Ni is the number of the i in
transverse cracking according to the chain code of the the chain code of a cracking.
cracking. The chain code is only from the minimum to The equivalent width of the cracking is obtain by
maximum of the X or Y coordinate. For each chain code dividing the area of the cracking by its length.
of the cracking, this paper calculates percentage of the
chain code. If the 0 occurs more than the 1, then the type B. Alligator cracking
of the line cracking is longitudinal cracking, else the type
of the line cracking is transverse cracking. Most of the alligator cracking are irregular and this
kind of cracking must be dug and repaired with a new
rectangle in the pavement maintenance. Thus, the feature
of the alligator cracking is the area of its box rectangle.
The box rectangle of the alligator cracking is shown in

2-142
The Ninth International Conference on Electronic Measurement & Instruments ICEMI’2009
figure 11. The area can be calculated by the length [6] F.F. Liu, G.A. Xu, Y.X. Yang, X.X. Niu and Y.L. Pan, “Novel
Approach to Pavement Cracking Automatic Detection Based on
multiplying the width of the box rectangle. Segment Extending”, International Symposium on Knowledge
The Cracking Index(CI) of the pavement detected is Acquisition and Modeling. pp. 610-614, Dec. 2008
expressed as below: [7] J. Zhang, “Study on Pavement Crack Identification and
, in which A is Evaluation Technology Based on Digital Image Processing”,
Xi’an: Chang an University, 2004
the area of the pavement tested.

Fig. 11. The box rectangle of the alligator cracking

V. CONCLUSION

This paper uses digital image processing techniques to

process the pavement cracking images acquired by the
digital camera. After the preprocessing to the cracking
images, this paper uses the thresholding method to obtain
the binary cracking images, and then carries on
morphological filtering operations to the images to get
better binary images.
The pavement cracking are classified to line crackings
and alligator cracking according to their topological
properties, then the line crackings are classified to
longitudinal and transverse cracking according to the
chain code of the cracking in the image. At the same time,
the paper calculates the length and width of the line
crackings and the area of the box rectangle of the
alligator cracking. Finally, the paper calculates the
Cracking Index (CI) of the pavement.

REFERENCES

[1] N.T. Sy, M. Avila, S. Begot and J.C. Bardet, “Detection of defects
in road surface by a vision system”, The 14th IEEE
Mediterranean Electrotechnical Conference. pp. 847-851, May.
2008
[2] B.C. Sun and Y.J. Qiu, “Pavement Crack Diseases Recognition
Based on Image Processing Algorithm”, Journal of Highway and
Transportation Research and Development, Vol. 25, No. 2, pp.
64-68, Feb. 2008
[3] J. Zhang, A.M. Sha, Z.Y. Sun and H.G. Gao, “Pavement Crack
Automatic Recognition Based on Phase-grouping Method”,
China Journal of Highway and Transport, Vol. 21, No. 2, pp.
39-42, Mar. 2008
[4] M.D. Yan, S.B. Bo, K. Xu and Y.Y. He, “Pavement Crack
Detection and Analysis for High-grade Highway”. ICEMI ‘07.
8th International Conference on Electronic Measurement and
Instruments, 2007. pp. 4-548 - 4-552, July. 2007
[5] X.J. Yu, R.D. Ma, Y. Liu and D.F. Zhao, “Study of fractal theory
on image segmentation of pavement surface crack”,
Microcomputer Information, Vol. 24, No. 2-3, pp. 302-304, 2008

2-143