Professional Documents
Culture Documents
Abstract—According to the characteristics of pavement crack of image edge’s damage caused by the removal of noise.
images, this paper introduce mathematical morphology into the Enhance the connectivity of the fractures, sequentially further
applications of road image analysis and processing, in addition improve the system's detection accuracy.
propose a new method combined with the mathematical
morphology to detect the road crack image. Experimental results
show that the method can not only effectively suppress noise, but II. IMAGE REPAIR AND PRUNING TECHNIQUES BASED ON
also clearly and accurately extract the edge of the image, and MATHEMATICAL MORPHOLOGY
solve the problem of image edge’s damage caused by the removal As the measurement of crack length is more complex, we
of noise. Enhance the connectivity of the fractures, sequentially need to thinning the binary image split processed, so that
further improve the system's detection accuracy. cracks could be reduced to a single pixel wide line in a regular.
What is more, Crack images in the binary segmentation could
Keywords- Pavement detection; Mathematical morphology; damage the edge of cracks, causing cracks in the image
Dilation and erosion; Closing operation
unconnected, and making the length measurement has some
bias. Therefore, we can use mathematical morphology method
I. INTRODUCTION to further repair and pruning the images of poor connectivity to
Edge detection is the basic techniques in the field of image reduce the fracture characteristics measurement’s bias.
analysis, pros and cons of image edge detection algorithm Mathematical morphology is an emerging discipline in the
directly determine the accuracy of crack identification. field of image processing and pattern recognition, has a
Literature [l] use the traditional method of differential operator rigorous mathematical theory, which is first proposed by
(Roberts, Sobel, Laplacian, Canny criteria algorithm) to solve Matheron and Serra[5]. Its main research method is that
two-dimensional real function of the gradient, and then select processing image from the a collection, and analysis image
the appropriate threshold to extract the image edges. For the based on structural elements. The basic idea is to use a certain
noise and different kind of crack disease’s pixel distribution in form of structural elements to measure and extract the image’s
the pavement images, Literature [2] proposes a multi-faceted corresponding shape, in order to achieve the purpose of image
structure element morphology edge detection algorithm, which analysis and recognition[6]. Mathematical morphology
compared to the traditional edge detection has a strong anti- algorithms take morphological analysis and processing
noise ability. Literatures [3-4] apply the traditional methods of algorithms in parallel, greatly improving the image analysis
mathematical morphology to the edge detection effectively and and processing speed.
filter out the noise at the same time without pretreatment.
Although the road crack detection has been studied in many Depending on the types of image processed, mathematical
documents, but it can not fully meet the need of the rapid morphology can be divided into binary morphology and gray-
development of highways. Mainly because of the noise scale morphology, the seven basic transformations is included
removal effect is not obvious, the measurement of fracture in which, swelling, erosion, opening, closing, hit, thickness and
characteristics is affected or more complex to implement. thinning. Because after segmentation the image is a binary
image, so this paper is mainly used binary morphological
According to the characteristics of pavement crack images, approach. Let the collection of binary images is A, the
this paper introduce mathematical morphology into the collection of structural elements is B, then:
applications of road image analysis and processing, in addition
proposes a new method combined with the mathematical Dilation is defined as:
morphology to detect the road crack image. This method is
mainly to further repair and pruning for pre-segmentation crack
{ }
road image, improve the accuracy of follow-up cracks feature
vectors extracted. Experimental results show that the method ⎣ x ( )
A⊕ B = x | ⎡ B ∩ A⎤ ≠ ∅
⎦ (1)
can not only effectively suppress noise, but also clearly and
accurately extract the edge of the image, and solve the problem Erosion operation is defined as:
968
Authorized licensed use limited to: Ming Chi University of Technology. Downloaded on July 27,2021 at 15:54:28 UTC from IEEE Xplore. Restrictions apply.
Figure 2. A series of morphological process results.
IV. CONCLUSION
This paper presents a new method combined with the
mathematical morphology to detect the road crack image. This
method is mainly based on the characteristics of
mathematicalmorphology operations for further repair and
pruning the pre-processed image. It not only effectively
suppresses the noise, but also clearly and accurately extracts
the edges of cracks. At the same time, it improves the accuracy
of cracks area and length measurement and provides a high-
quality guarantee for the follow-up cracks’ accurate location.
These methods and techniques have wide application prospects
in pavement automatic detection and maintenance systems, for
similar projects of image process also have a reference.
ACKNOWLEDGMENT
This work is supported by the Chongqing Natural Science
Foundation (CSTC, 2010AC6074) and Chongqing Jiaotong
University Graduate Innovation Foundation.
REFERENCES
[1] Kelvin, “Designs and implementation of automated systems for
pavement surface distress survey,” Journal of Infrastructure Systems, vol.
3, pp. 24-32, 2000.
[2] LI Gang, HE Yuyao, “Edge detection for road crack image with
multidirection morphological structuring elements,” Computer
Engineering and Applications, vol. 46, pp. 224-226, 2010.
[3] Guo Xiaoxin, Xu Zhiwen,Pang Yunjie, “An adaptive soft morphological
gradient filter for edge detection,” IEEE Proceedings of the 3rd
International Conference on Image and Graphics, 2004.
[4] Mukhopadhyay S, Chanda B, “Multiscale morphological segmentation
of gray-scale images,” IEEE Transactions on Image Processing, vol. 12,
pp. 533-549, 2003.
[5] Abe S, Bourland H, Adali T, Bengio S, “Analysis of support vector
machin,” Proceedings of IEEE Signal Processing Society Workshop,
2002.
[6] Lee J S J, Haralick R M, “Morphology edge detection,” IEEE Transon
Robotics Automat, pp. 410- 156, 1987.
969
Authorized licensed use limited to: Ming Chi University of Technology. Downloaded on July 27,2021 at 15:54:28 UTC from IEEE Xplore. Restrictions apply.