DOI: 10.1784/insi.2011.53.10.
552 AEROSPACE RADIOGRAPHY
Crack detection using image processing techniques for
radiographic inspection of aircraft wing spar
B Stephen Wong, Xin Wang, Chen Ming Koh, Chen Guan Tui, ChingSeong Tan and Jian Xu
Radiographic inspection is one of the most important
non-destructive inspection methods for aircraft wing spar
inspection. Traditionally, radiographic inspection is time
and manpower consuming. In addition, human inspection
of cracks and damage based on film radiography is very
subjective, inconsistent and sometimes biased. To improve
the speed and efficiency of radiographic inspection, digital
X-ray systems are gaining popularity over traditional
film-based X-ray. Digital radiography has emerged as a
leading technology for aircraft inspection. Therefore, it is
desirable to develop a computer-aided system to assist in
the interpretation of digital radiographic images to increase
the objectivity, accuracy and efficiency of radiographic
interpretation. In this paper, a crack detection algorithm
using region of interest (ROI) minimisation, thresholding
and edge detection techniques is developed for radiographic
inspection of an aircraft wing spar.
1. Introduction
The use of non-destructive testing (NDT) in the inspection
of aircraft parts and components is crucial in determining the
structural integrity of aircraft. Commonly used NDT methods
in aerospace are visual inspection, liquid penetrant inspection,
magnetic particle inspection, eddy current inspection, ultrasonic
inspection, radiographic inspection and leak testing[1]. Some
advanced NDT techniques such as acousto-ultrasonic techniques,
ultrasonic phased array and magnetic sensors can also be applied in
aircraft inspection[2-7].
Defects on the aircraft wing are usually non-visible. Traditional
radiography is one of the effective NDT techniques used for the
detection of cracks and damage in aircraft structures[8]. To improve
the speed and efficiency of radiographic inspection, digital X-ray
systems are gaining popularity over traditional film-based X-ray.
Digital radiography makes use of digital X-ray sensors to convert
the X-rays into visible images. Digital radiography has emerged
as a leading technology for aircraft crack inspection. It offers
many advantages over conventional film-based radiography[9,10].
Digital radiography can not only save the time spent on chemically
B Stephen Wong and Chen Ming Koh are with the School of Mechanical
and Aerospace Engineering, Nanyang Technological University (NTU),
Singapore.
Xin Wang* is with the Faculty of Engineering and Science, Universiti Tunku
Abdul Rahman (UTAR), Malaysia.
Chen Guan Tui is with the Republic of Singapore Air Force (RSAF),
Singapore.
ChingSeong Tan is with the Multimedia University (MMU), Malaysia.
Jian Xu is with the Singapore Institute of Manufacturing Technology
(SIMTech), Singapore.
*Corresponding author. Email: wangx@utar.edu.my
552
Submitted 19.07.11
Accepted 19.09.11
developing films, it also allows images to be stored digitally. To
fully utilise the benefits of digital radiography and further improve
the inspection process, an automatic crack detection software is
required to aid inspectors to quickly recognise crack features from
digital radiographs.
In our previous work[11], we identified suitable digital X-ray
detectors to meet the high sensitivity required to inspect the aircraft
wing. Different sensors were evaluated by using a set-up of a wire
IQI penetrameter, aluminium specimens with 6 mm thickness and
a 2T-hole penetrameter. We recommended amorphous silicon flat
panels to be the most suitable detector for the aircraft inspection
applications. We also developed a crack detection algorithm using
region of interest (ROI) minimisation and improved Canny edge
detection techniques for aircraft wing inspection[12]. However,
the algorithm has limited accuracy when tested on aircraft wing
spar images that are noisy or cluttered with many features. In this
paper, an improved algorithm is developed for aircraft wing spar
radiographic inspection.
2. Robot guided X-ray system for aircraft
wing inspection
A robot guided digital X-ray imaging system, as shown in Figure
1, is developed for automatic aircraft wing inspection. The
manipulator system automatically moves an X-ray tube and a
digital sensor along the wing of the aircraft. Digital radiographs are
taken as the system moves and the whole wing area can be built up
gradually. The movement is programmed from a database drawing
of the aircraft wing.
Figure 1. Robot guided scanning manipulator
Figure 2 is the aircraft wing spar radiograph. In this radiograph,
line cracks and fatigue cracks created on specimens are taken with
a wing spar image, with the cracks circled in yellow. Even though
the cracks are visible to the human eye, it takes close and careful
observations to detect them. Moreover, there are numerous rivet
holes in each radiograph and these areas are the high-risk regions
for fatigue cracks, which require more attention. Having numerous
Insight Vol 53 No 10 October 2011
Figure 2. Typical aircraft wing spar radiograph

rivet holes that inspectors are required to scrutinise closely puts

them under intense stress and hinders their efficiency. With a good
crack detection algorithm that focuses around the rivet holes,
inspectors will be able to make quick and accurate conclusions on
the aircraft wing spar radiographs.

3. Crack detection algorithm

Firstly, we tested the algorithm developed by Wang et al[12] using
some aircraft wing spar images. The presence of noise is either
mistaken as cracks, or drowns out the crack features from the
algorithm (Figure 3), thus some improvement is needed to increase
the accuracy of the algorithm. In addition to the algorithm, which
recognises cracks through grey-level gradients and the dimensions Figure 3. Detection results using Wang et al method
of the features detected, other parameters need to be considered to
make crack detection more accurate.
Suggested improvements include the removal of rivet holes
from the region of interest, as well as using thresholds instead of
edge detection to find rivet holes.
3.1 Improvement on region of interest (ROI) generation
As seen in the test on image No 2, as shown in Figure 3, there
is a wrong detection marked within the rivet holes. However, the
likeliness of a crack occurring on a rivet is very low, thus the need
to detect cracks within the rivet area is questionable. The edge of a
rivet hole is also sometimes mistaken as a crack, as seen from image
No 2. It is evident that the rivet itself is causing some problems
for the accuracy of detection. In the test on image No 4, the crack
detected in the ROI is grouped as a single object with the rivet
hole itself, thus being excluded after shape recognition processes.
This further compounds the problem of having the rivet hole’s area
included in the ROI as it clearly interferes with the crack.
To solve the problem of the rivet hole interfering with crack
recognition, the rivet hole is removed from the ROI, leaving only
the area around the rivet.
Figure 4 shows the ROI obtained when the new algorithm is Figure 4. (a) Binary image of holes in test image No 1; (b)
used on test image No 1. Figure 4(a) is the binary image of the expanded mask area; (c) slightly expanded mask area; (d) final
holes detected, Figure 4(b) is the enlarged masking area and Figure ROI
4(c) is the slightly increased masking area to cover the edges of
the holes. The final ROI is obtained by subtracting the result of image edge matrix. The result of this final operation on test image
the second from the first, and the result can be seen in Figure 4(d). No 1 is shown in Figure 5(a). Figure 5(b) shows the final crack
After obtaining the ROI, the edge information within the ROI is detected on test image No 1 when the new algorithm is included,
obtained using element-by-element multiplication with the full where all cracks are detected instead of four out of five previously.

Insight Vol 53 No 10 October 2011 553


Figure 5. (a) Edges detected within new ROI of test image No 1;
(b) final detection with new algorithm
3.2 Hole detection by thresholding
In image No 1 of Figure 3, one of the cracks is not detected. On
closer inspection this is because the undetected crack is recognised
as a single object together with the hole. This results in the crack
being filtered out together with the hole as they did not have a shape
that resembled a crack. Figure 6 shows how the edge detected crack
and hole are joined as one entity.
Figure 6. ROI of image No 1 in Figure 3 shows crack and hole
grouped as the same object
Figure 7 shows the result of image No 2 in Figure 3. There are
four wrong detections which are marked by yellow circles, with
three of them on the rivet hole and identifying the rivet as a crack.
Another wrong detection marks the edge of a specimen hole as a
crack. The white lines are IQI wires which were added and meant
to simulate cracks, so their detection is considered acceptable. The
dark coloured hole, which is the specimen hole, is not detected.
Figure 8 shows the possible hole features after edge detection. It
can be seen that the dark specimen hole is not fully detected by
the Canny edge function. As a result, the hole is not an enclosed
object and cannot be filled to form a masking area. As a result, a
Figure 7. Detected cracks in image No 2
554
crack from the specimen is not detected. The edge of the specimen
hole is, however, in the ROI of the true rivet hole, and is mistaken
as a crack. Image No 3 in Figure 3 is the same as image No 2 but
has a different implanted crack and therefore has a similar source
of error.
Figure 8. Edge detected possible holes in test image No 2
In image No 4 of Figure 3, the algorithm fails entirely because
the rivet hole is not detected. Failure to detect rivet holes results in
the failure to set a ROI, thus nothing will be detected. The reason
for failing to detect rivet holes is the same as the reason specimen
holes could not be detected in image No 2 and No 3, and that is
because the edge detected rivet hole is not complete to form an
enclosed object.
The tests conducted on image No 2, No 3 and No 4 showed
that Wang’s algorithm has problems detecting rivet holes when
edge detection could not form an enclosed object on the holes. This
means that even the slightest failure of edge detection in detecting
a small part of the hole will lead to the entire hole being neglected,
thus missing any possible cracks around such holes.
The authors therefore propose an alternative method of detecting
the rivet holes which is less punishing on imperfect accuracy, and
that is by thresholding instead of by edge detection. Thresholding
is the simplest method of image segmentation. From a greyscale
image, thresholding can be used to create binary images[13]. In
image thresholding, a range of grey level is selected and only pixels
that satisfy the grey levels are selected. Therefore, this method is
very useful in images where the relevant objects have a distinct
difference in grey levels to the background.
The new algorithm requires a threshold value to be explicitly
defined first, 110 for the case shown. If the rivet holes are bright
pixels, they will be selected by picking pixels in the original image
that are more than the threshold. If the hole pixels are dark, then
they will be selected by calling for pixels less than the threshold.
The rest of the algorithm is similar to Wang’s algorithm, which
performs recognition of the rivet holes by their area and eccentricity.
Figure 9(a) shows the effect of using thresholding to find rivet
holes in test image No 4. The holes can be found, but together with
some IQI wires that are meant to simulate cracks. The algorithm is
able to make detections in the end, as shown in Figure 9(b).
The flow chart of the new algorithm is shown in Figure 10 and
the detection results using the new algorithm are shown in Figure 11.
4. Experimental results
The new algorithm was tested on eight images, as shown in Figure
12.
Table 1 shows the detection results using the original algorithm.
Out of a total of 18 cracks in the tested images, 14 cracks were
detected with four wrong detections. Table 2 shows the detection
results using the new algorithm. In general, the new algorithm
is able to detect slightly more cracks with no wrong detections.
Out of a total of 18 cracks in the tested images, 16 cracks were
Insight Vol 53 No 10 October 2011
Figure 9. (a) Result of thresholding on test image No 4; (b) final
detection on test image No 4 with new algorithm

Figure 11. Detection results using new algorithm

Table 2. Results of image test with new algorithms incorporated

Image Number of Correctly Non- Cracks Minimum

No cracks detected cracks detected detectable
cracks detected as non- crack length
Figure 10. Flow chart of new algorithm as cracks cracks (pixels)
1 1 1 0 0 25
detected without a wrong detection. The new algorithm improves 2 2 2 0 0 11
the accuracy of aircraft wing spar inspection. 88.9% of cracks
are detected using the new algorithm, while 77.8% of cracks are 3 4 4 0 0 8
detected using the old algorithm. 4 1 1 0 0 28
Table 1. Results of image test using original algorithm 5 5 5 0 0 22
6 2 1 0 1 25
Image Number Correctly Non- Cracks Minimum
No of cracks detected cracks detected detectable 7 2 1 0 1 21
cracks detected as non- crack length 8 1 1 0 0 18
as cracks cracks (pixels)
Total 18 16 0 2
1 1 1 0 0 25
2 2 2 0 0 11
3 4 4 0 0 8
5. Conclusion
4 1 1 0 0 28 The development of automated crack detection software has a
huge importance in improving the efficiency and accuracy of
5 5 4 0 1 22
radiographic inspection. Currently, the developed algorithm
6 2 1 4 1 25 can perform well in the localised images tested. However, good
7 2 1 0 1 21 performance of current algorithm requires manual tweaking of
8 1 0 0 1 NA certain parameters, thus it still lacks complete automation. The
crack detection software is a function of the radiographs and cracks
Total 18 14 4 4 it would encounter. The recommendations for future developments

Insight Vol 53 No 10 October 2011 555

 also depend hugely on the accuracy of maintaining conditions in
X-ray shoots. Therefore, further developments will be dependent
greatly on changes in the quality of digital radiographs, as well as
a good number of wing radiographs with cracks present to base
future development on.
Acknowledgements
The authors gratefully acknowledge the support from Nanyang
Technological University, Republic of Singapore Air Force and
Singapore Technology Aerospace.
References
1. R Prasad and N S L Ganesh, ‘Defect analysis in aerospace
systems using non-destructive testing’, Proc National
Seminar on Non-Destructive Evaluation, Hyderabad,
December 2006.
2. N Guo and P Cawley, ‘Lamb wave propagation in composite
laminates and its relationship with acousto-ultrsonics’,
NDT&E International, 26 (2), pp 75-84, 1993.
3. L P Scudder, D A Hutchins and N Guo, ‘Laser-generated
ultrasonic guided wave in fibre-reinforced plates – theory and
experiment’, IEEE Transactions on Ultrasonics, Ferroelectrics
and Frequency Control, 43 (5), pp 870-880, 1996.
4. D J Roth, R P Tokars, R E Martin, R W Rauser, J C Aldrin and
E J Schumacher, ‘Ultrasonic phased array inspection
simulations of welded components at NASA’, Materials
Evaluation, Vol 67, No 1, January 2009.
5. S Mahaut, S Chatillon, E Kerbrat, J Porré, P Calmon and
O Roy, ‘New features for phased array techniques inspections:
simulation and experiments’, 16th World Conference on Non-
Destructive Testing, Montréal, 2004.
6. B Lebrun, Y Jayet and J C Baboux, ‘Pulsed eddy current
application to the detection of deep cracks’, Materials
Evaluation, Vol 53, No 11, pp 1296-1300, 1995.
7. W F Avrin, ‘Eddy current measurements with magnetoresistive
sensors: third-layer flaw detection in a wingsplice structure
25 mm thick’, Proceedings of SPIE, Vol 3994, pp 29-36,
2000.
8. R D Bowman, B A Bennett and M E Stevenson, ‘Radiographic
inspection in failure investigations’, Journal of Failure Analysis
and Prevention, 3 (3), pp 73-77, 2003.
9. R Halmshaw, Industrial Radiology: Theory and Practice,
Second Edition, Chapman and Hall, UK, 1995.
10. U Ewert, U Zscherpel and K Bavendiek, ‘Replacement of film
radiography by digital techniques and enhancement of image
quality’, Online NDT Journal, Vol 12, No 6, 2007.
11. X Wang, B Stephen Wong, C G Tui, K P Khoo and F Foo,
‘Real-time radiographic non-destructive inspection for aircraft
maintenance’, 17th World Conference on Non-Destructive
Testing, Shanghai, China, 25-28 October 2008.
12. X Wang, B S Wong, C S Tan and C G Tui, ‘Automated crack
detection for digital radiography aircraft wing skin inspection’,
Research in Nondestructive Evaluation, pp 105-127, 2011.
13. L G Shapiro and G C Stockman, Computer Vision, Prentice
Figure 12. Test new algorithm Hall, 2002.

556 Insight Vol 53 No 10 October 2011