Available online at www.sciencedirect.

com

ScienceDirect
Aquatic Procedia 4 (2015) 522 – 531

INTERNATIONAL CONFERENCE ON WATER RESOURCES, COASTAL AND OCEAN

ENGINEERING (ICWRCOE 2015)

Wave Flume Measurement using Image Analysis

K. Viriyakijjaa* and c. Chinnarasria
a
Water Resources Engineering & Management Research Center (WAREE)
King Mongkut’s University of Technology Thonburi, Bangkok, Thailand, 10140

Abstract

Wave observation represented water surface elevation at a reference point is basic and essential data for ocean and coastal
engineering that is mostly measured with traditional wave gauges, for instance capacitance, resistance and conductivity types.
Therefore, accuracy and exactness are the most important aspects of data acquisition systems. In this study, we focus on the
development of image processing techniques for wave measurement in a wave flume as a tool to replace traditional wave gauges,
which are complicated to use, costly and intrusive. Recently, image processing techniques have been developed for utilization in
various tasks. The main factors that influence the accuracy of these techniques are image quality, algorithm performance and
exact calibration. Image processing techniques based on edge detection algorithms were used to measure waves in the laboratory
with a camera. The experimental results showed the agreement of water surface elevation from both measurement methods. The
average wave height and average wave period of regular wave obtained from the zero up-crossing analysis were different by less
than 10% and 1% respectively compared with traditional wave gauges. Moreover, this algorithm can also measure irregular
waves in hydraulic laboratories as well. Hence, this approach is a new, alternative tool for use in wave flume measurement, with
good results.
© 2015 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license
© 2015 The Authors. Published by Elsevier B.V.
(http://creativecommons.org/licenses/by-nc-nd/4.0/).
Peer-review under
Peer-review under responsibility
responsibility of organizing
of organizing committee
committee of ICWRCOE
of ICWRCOE 2015 2015.

Keywords: Image processing, Edge detection algorithms, Regular wave, Zero up-crossing, Irregular wave

1. Introduction

In general, wave characteristics play an important role in ocean and coastal engineering research, and are used to
improve overall understanding of wave phenomena. One example of this is the role of coastal erosion. The main
parameters of waves are wave height, wave period and wave length, which are obtained by analysis of continuously

* Corresponding author. Tel.: +662-470-9132; fax: +662-427-9063.

E-mail address: [email protected]

2214-241X © 2015 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license
(http://creativecommons.org/licenses/by-nc-nd/4.0/).
Peer-review under responsibility of organizing committee of ICWRCOE 2015
doi:10.1016/j.aqpro.2015.02.068
 K. Viriyakijja and C. Chinnarasri / Aquatic Procedia 4 (2015) 522 – 531 523

water surface elevations with selected time intervals at an interesting point. Various observation techniques were
used in field measurement of coastal and ocean study, for example marine radar, satellite imaging, airborne LiDAR
survey and small man-made aerial vehicle (de Vries et al., 2011; Vousdoukas et al., 2014). Due to the complications
of field study, the wave flume was simulated so that it could be studied for any cases like a common test facility.
Mostly researchers monitor wave mechanisms with traditional wave gauges such as capacitance, resistance and
conductivity types in their experimental work. However, some investigators proposed alternative methods to be used
as an equally one as traditional tool in wave flume, which are laser scanning, computer vision, LiDAR systems,
ultrasonic sensor, video and image processing techniques. Knowledge of image processing techniques has been
carried out in different fields over decades such as sorting objects, vehicle speed detection and finding defects in
electronic parts. In coastal and ocean laboratories, image processing from a charge-coupled device (CCD) video
camera is still an interesting approach for application in wave flume measurement. Nevertheless, the significant
aspects not only the obtained data and analysis process but also the accuracy and reliability of this approach.
Over the past two decades, a few researchers have attempted to measure surface wave profiles by image analysis
with complicated calculation processes (Bonmarin et al., 1989). The monitoring of swash zone hydrodynamics was
experimentally studied using video techniques applied with ArcView (Foote and Horn, 1999). The wave profile was
analysed in a wave flume facility by the Mexican Hat wavelet transform and GVF snake image-based techniques
(Lee and Kwon, 2003; Yao and Wu, 2005). Computer vision was used to detect the water surface in a wave flume by
side wall observation (Erikson and Hanson, 2005; Iglesias et al., 2009). Kuo et al. (2009) studied the separate
incident and reflected wave method through submerged breakwater with image processing using a CCD video
camera in the laboratory. Almar et al. (2012) proposed the direct breaker height estimator from the video that
processed along cross shore of basin experimental study. There have been several efforts to monitor both water
surface elevation and bed topography. Wang et al. (2012) developed the CCD camera based on the modified plane-
based camera calibration method (MPCC) to investigate the sand bed and waveform measurement. The laboratory
investigation of sand bed evolution was carried out in a nearshore zone by using video camera detection (Foti et al.,
2011; Astruc et al., 2012; Palmsten and Holman, 2012). Chinnarasri and Viriyakijja (2014) applied the image
processing method to experimentally investigate the performance of a rectangular box type of single and double
floating breakwater in a wave flume.
The details of previous studies showed that there are some reference works on a similar concept. Hence, this
paper proposes alternative techniques to be utilized for ocean and coastal observation systems in a hydraulics
laboratory. In the present study, the CCD video camera is experimentally applied to use as a traditional wave gauge
in the wave flume measurement using image processing techniques based on edge detection algorithms.

2. Experimental set-up

2.1. Observation systems

The CCD video camera (AVTECH IP Camera) with a frame rate of 30 frames per second was mainly used in the
observation systems. The scene of interesting area was recorded witha sensor resolution of 1280 u 720 dpi. Two
CCD video cameras were placed at the exactly perpendicular positions to the side window of the wave flume. To
compare the wave series data measured from these CCD video cameras, two capacitance wave gauges (KENNEK
CH-403A and CHT4-40) were used for validation in the same position asthe wave flume with a sampling rate of 50
Hz.

2.2. Test facilities

The experimental studies were carried out in two wave flumes. The specific details are described in Table 1.
First, the image processing technique was validated with a capacitance wave gauge in a narrow wave flume in the
hydraulics and coastal laboratory of Chulalongkorn University. Second, the image processing was applied for use as
a tool for measuring waves passing through the floating breakwater in a wide wave flume at the Civil Engineering
Department of King Mongkut’s University of Technology Thonburi.
524 K. Viriyakijja and C. Chinnarasri / Aquatic Procedia 4 (2015) 522 – 531

Table 1. Details of test facilities

Properties Narrow wave flume Wide wave flume
Width of wave flume (m) 0.60 1.0
Length of wave flume (m) 16.0 27.0
Depth of wave flume (m) 0.75 1.0
Still water level (m) 0.30 0.60
Material of bed flume Steel Steel
Material of side flume Acrylic sheet Glass
Wave generator type Flap type with AC motor Piston type with DC motor
Wave condition Regular wave Irregular wave

2.3. Physical model

In order to test the structure under irregular wave action, the floating breakwater was used to study the
attenuation of wave height in a wide wave flume as shown in Fig.1. This module is an aluminium box in shape that
consists of two parts, i.e. the main box and side plates. The floating breakwater position was 6.50 m away from the
wave generator. The structure was restrained with a heavy steel plate at the bed flume by a steel cable. The clearance
between the model and side wall of the wave flume was fixed at 0.25 m.

0.50 m 0.50 m
0.15 m

0.30 m

Fig. 1. Physical model of floating breakwater.

2.4. The experimental methods

The methodology of this study wasto use a video camera as a versatile instrument in wave flume measurement.
The step-by-step process was as follows. First of all, the point of water surface elevation observation was defined,
then a scaling mark was placed on the side flume at the investigation point. Secondly, video cameras were mounted
at a suitable distance away from the target plane depending on the desired data resolution. Thirdly, the frame of
recording was arranged into the line of the still water level and the camera level was adjusted both vertically and
horizontally. Fourthly, a uniformly solid background was set behind the video camera to block the reflection shadow
from unwanted objects. Finally, the video cameras started to record with varying water surface elevations.
 K. Viriyakijja and C. Chinnarasri / Aquatic Procedia 4 (2015) 522 – 531 525

3. Image processing techniques

Nowadays, image processing techniques are used widely in many studies. The main concept of image processing
techniques is the transformation of image data into digital form (Erikson and Hanson, 2005). Therefore, the
accuracy of this method is the most important aspect and consists of three factors: image quality, algorithm
performance and exact calibration (Wang et al., 2012).

3.1. Image quality

Noise, or an unwanted or disturbed signal in the image, is the main factor that causes low image quality. In this
experiment, there are many ways to deal with noise reduction,such as the following:

x The CCD video camera was mounted about 1.20 m away from the clear acrylic side wall of the wave flume for
appreciated image boundary.
x To create the obvious contrast between water and background, the opposite side wall was fabricated from dark
blue steel plate.
x A solid homogeneousboard was placed behind the CCD camera to deal with reflected shadow in the frame of the
monitoring area.
x The light around the laboratory area was controlled to reduce noise.

3.2. Performance of algorithm

Edge detection is one of the techniques used normally for image processing and many edge detectors are
available today, such as Roberts, Sobel, Prewitt, Laplacian and Canny operators. The Canny edge detector, invented
by John Canny in 1986, was used in MATLAB for wave observation since it is an optimal, powerful and effective
edge detector (Gonzalez and Woods, 2002; Jain et al., 1995; Marques, 2011; McAndrew, 2004). The basic steps are
divided into four parts, as follows:

x Smoothing with a Gaussian filter: filters out noise from the image by Gaussian filter and then the potential edges
are selected as a candidate image pixel.
x Gradient analysis: calculates the magnitude and direction of the gradient by a first-order derivative method for
edge strength.
x Non-maximum suppression: thins the edge by non-maximum suppression leaving only the pixel at the top of
each edge.
x Hysteresis thresholding: identifies the edge pixel position and connects the edge boundary by hysteresis
thresholding.

3.3. Exact calibration

Video cameras are base apparatus for laboratory measurement. The cameras setting is shown in Fig. 2. Most
camera devices suffer from lens distortion or optical error that cause straight lines in the real world to become
curved lines in the image (Foote and Horn, 1999; Erikson and Hanson, 2005). In this experimental investigation,
three approaches were used to confirm exact calibration.

x Using an array of exact grids made of clear acrylic sheets to check the lens distortion in each camera position.
x Mounting the camera to record data in line with the still water level in the wave flume and controlling the tilt
angle of the camera to be vertically and horizontally level.
x Designing the algorithm to process only the middle area of the image since lens distortion barely occurs in the
middle of an image.
526 K. Viriyakijja and C. Chinnarasri / Aquatic Procedia 4 (2015) 522 – 531

1.20 m

Fig. 2. The video cameras setting.

4. Edge detection algorithms

A wave is a continuously changing water surface elevation over time (Hughes, 1993; Kamphuis, 2000).
Therefore, the concept of designing algorithms to measure waves refers to the transferring of the time domain
represented by many video image frames to the frequency domain represented by water surface elevation. Edge
detection algorithms were used for detecting the water surface as an edge in each frame and converting the pixels
between the edge and reference point into water surface elevation. The results of edge detection algorithms are
shown in Fig. 3. In this study, MATLAB was used to design these algorithms and the details of the algorithms in
each step are divided into three processes as follows.

Colour picture Greyscale picture Edge picture

Fig. 3. Results of edge detection algorithms.

4.1. Preprocessing

First, the video of water surface elevation monitoring at an interesting point was extracted to the sequence of
image frames. Secondly, all of the frames were defined in a specific area at thecentre of each frame to decrease the
analysed time. Furthermore, this step partly reduced lens distortion since the distortion occurs in the middle of
frames less frequently than in the border area of the image frame. Thirdly, the image frames were converted from
RGB (three channels) to greyscale (one channel) images for effective processing.

4.2. Processing

The greyscale images were analysed with a Canny edge detector in MATLAB. At the appropriate threshold, a
good binary image was obtained. The binary image was composed of two parts: 1: the foreground or edge
represented by the colour white (pixel value = 1), and 2:the background represented by the colourblack (pixel value
 K. Viriyakijja and C. Chinnarasri / Aquatic Procedia 4 (2015) 522 – 531 527

= 0). After that, the binary image was processed in the algorithm by detecting the edge or water surface elevation
and showed the result by referring to the exact scale at the image. One frame was one point of water surface
elevation, therefore the sequence of image frames was equal to the continuously changing water surface elevation
over time.

4.3. Post-processing

The final result from the edge detection algorithm was the relationship of water surface elevation at the
monitoring position. Then the main wave characteristics, represented by wave height, wave period and wave length,
were acquired by using zero up-crossing analysis and Hunt’s empirical equation.

5. Wave analysis

The videos of water surface elevation were obtained by filming from CCD video cameras, which were used to
analyse edge detection algorithms in MATLAB. The result of MATLAB is the continuous water surface elevations
at interesting time intervals. This output was used to find out the wave characteristics by short-term analysis and an
empirical equation, respectively.
Zero up-crossing is a common analysis method used for short-term wave analysis. The relation of water surface
elevation with time intervals was resolved with this technique to obtain wave height and wave period (Kamphuis,
2000; Holthuijsen, 2007). The process of zero up-crossing analysis can be described as follows:

x Define the reference point represented by the solid point in Fig. 4 that passes from a negative to a positive
position of water surface elevation and intercepts with the reference line.
x The wave height of the first loop is the maximum difference of water surface elevation between two solid points
in Fig. 4.
x The wave period of the first loop is the time interval from the first solid point to the second one in Fig. 4.
x These three steps along the reference line should be carried out to get the wave height and wave period of all
wave loops.

Fig. 4. Zero up-crossing analysis method.

6. Results and discussion

The experimental results obtained in 11 cases of both regular wave and irregular wave tests in two wave flumes
are shown in Tables 2 and 3, respectively. The observation of water surface elevations obtained from the CCD video
cameras were compared with those obtained from two capacitance wave gauges. The five cases of regular waves in
528 K. Viriyakijja and C. Chinnarasri / Aquatic Procedia 4 (2015) 522 – 531

a narrow wave flume were used for validation without structure. The other six samples of irregular waves were
investigated in the wide wave flume to test the image processing technique for the case of an existing floating
breakwater. All the irregular waves were generated by a manually controlled servo motor through the Mach3 CNC
controller program.
All of the regular wave cases were carried out in the narrow wave flume with increasing stroke at the same
frequency as the wave generator. The validation results showed that the difference in the average wave height of two
measurements is less than 10%. On the other hand, the average wave period rarely has different values, as can be
seen in Table 2.
In the case of regular waves, the comparison of water surface elevation shown in Fig. 5 indicates that some errors
occur in both measurement methods. In my view, the causes of error are:

x Difference in frequency: The different frequency rate of data observation between the CCD video camera (30 Hz)
and the capacitance wave gauge (50Hz). Hence, the obtained data are different interval timesover the
experimental works.
x Difference in point: The capacitance wave gauge was used to measure the water profile inside the wave flume
while the CCD video camera was mounted to detect the water profile outside the wave flume.

Table 2. Validation of regular wave characteristics between wave gauge and CCD video camera measurement
Wave gauge CCD video camera Difference
Case study Average Average Average Average Average wave Average wave
wave height wave period wave height wave period height (%) period (%)
(cm) (s) (cm) (s)
Regular wave 01(stroke = 2 cm) 2.55 0.818 2.38 0.822 6.67 0.49
Regular wave 02(stroke = 3 cm) 2.87 0.817 2.78 0.821 3.14 0.49
Regular wave 03(stroke = 4 cm) 3.67 0.818 3.67 0.816 0.00 0.24
Regular wave 04(stroke = 5 cm) 4.86 0.820 4.41 0.822 9.26 0.24
Regular wave 05(stroke = 6 cm) 6.08 0.828 5.67 0.829 6.74 0.12

Fig. 5.Water surface elevation in the case of regular wave 05.

Afterward, the irregular waves were generated in a wide wave flume to create the scenario of the real sea with
structure. Then the edge detection algorithm was used to observe the water surface elevation in the same way as
regular wave cases. The monitoring points were behind and in front of the floating breakwater to investigate the
attenuation of wave height. The representative wave heights of irregular wave cases are presented in Table 3.
In the case of real sea, the water surface elevation of the irregular wave case was experimentally obtained as
shown in Fig. 6 and Fig. 7, respectively. Because the physical floating breakwater was made from an impervious
material, the wave height decreased significantly through the aluminium floating breakwater box. This can be
 K. Viriyakijja and C. Chinnarasri / Aquatic Procedia 4 (2015) 522 – 531 529

explained that the mechanism of attenuated wave height is the reflection wave influenced by the floating
breakwater. Although using CCD video cameras for wave flume measurement should give more attention to
environmental conditions, this approach provides more advantages than traditional wave gauges. For more
information see Table 4.

Table 3. Application of irregular wave measurement with floating breakwater

Average wave height (cm) Significant wave height (cm)
Case study In front of Behind In front of Behind
floating floating floating floating
breakwater breakwater breakwater breakwater
Irregular wave 01 5.48 3.54 8.63 5.87
Irregular wave 02 4.37 3.79 6.70 5.95
Irregular wave 03 3.89 3.18 6.02 5.04
Irregular wave 04 7.22 3.91 8.56 6.46
Irregular wave 05 12.25 10.79 16.59 14.13
Irregular wave 06 6.82 4.40 11.18 7.56

Fig. 6. Water surface elevation in the case of irregular wave 02 (in front of floating breakwater).

Fig. 7. Water surface elevation in the case of irregular wave 02 (behind floating breakwater).
530 K. Viriyakijja and C. Chinnarasri / Aquatic Procedia 4 (2015) 522 – 531

Table 4. Comparison of wave gauge and CCD video cameras

Factors Wave gauge CCD video cameras
1) Cost and set-up operation High price and complicated Cheaper than traditional wave
systems gauge and easy to setup
2) Type of measurement Intrusive tool Non-intrusive tool and one
CCD video camera can work as
an array of traditional wave
gauges
3) Attention to environmental Less attention Pay more attention such as light
conditions control, reflected shadow and
lens distortion

7. Conclusions

The observation systems of CCD video cameras with edge detection algorithms are an innovative concept that
measures water surface elevation. This alternative technique has many advantages: for example, it is easy to setup,
low-cost, compared to an array of sensors, and provides good accuracy. The experimental results indicated that the
validation between CCD video camera and traditional wave gauge has a good agreement in terms of water surface
elevation,whereas the average wave height and average wave period of regular waves obtained from the zero up-
crossing analysis method showed differences of less than 10% and 1% respectively. This approach, moreover, can
also be appliedto irregular waves.
In future work, the image processing technique based on edge detection algorithms should be applied as an
alternative tool in a wave flume. More case studies are needed to investigate image processing techniques,which can
be used as efficiently as traditional wave gauges, such as wave run-up, sediment movement and seabed profile.
Furthermore, the improvement and development of this method for field observation are still interesting issues.

Acknowledgements

The authors would like to thank King Mongkut’s University of Technology Thonburi (KMUTT) for providing
research funding and test facilities under the National Research University Project. Also, sincere thanks go to Dr.
AnurakSriaiyawat for the opportunity to validate image processing techniques with wave gauges in a wave flume at
Chulalongkorn University and Dr. SupanneeTanathong for assistance with image processing techniques.

Appendix A. Hunt’s empirical equation

For more details of wave analysis, the wave celerity and wave length can be expressed as an empirical equation.
Hunt (1979) proposed a direct and accurate approximating equation for calculating wave length, which can be
expressed as Eq. [1]. This calculation approach is based on a higher-order expansion method that gives a highly
accurate result of 0.1% (Kamphuis, 2000).

C2 1
ª¬ y0  (1  0.6522 y0  0.4622 y02  0.0864 y04  0.0675 y05 )1 º¼ (1)
gd

2S d gT 2
Where y0 , L0 and L CT
L0 2S

Where C is the wave celerity, L is the wave length, L0 is the wave length in a deep-water zone, T is the wave
period, g is the acceleration of gravity and d is the still water level in the wave flume.
 K. Viriyakijja and C. Chinnarasri / Aquatic Procedia 4 (2015) 522 – 531 531

References

Almar, R., Cienfuegos, R., Catalán, P.A., Michallet, H., Castelle, B., Bonneton, P. and Marieu, V., 2012. A New Breaking Wave Height Direct
Estimator from Video Imagery. Coastal Engineering 61, 42-48.
Astruc, D., Cazin, S., Cid, E., Eiff, O., Lacaze, L., Robin, P., Toublanc, F. and Cáceres, I., 2012. A Stereoscopic Method for Rapid Monitoring of
the Spatio-Temporal Evolution of the Sand-Bed Elevation in the Swash Zone. Coastal Engineering 60, 11-20.
Bonmarin, P., Rochefort, R. and Bourguel, M., 1989. Surface Wave Profile Measurement by Image Analysis. Experiments in Fluids 7, 17-24.
Chinnarasri, C. and Viriyakijja, K., 2014. Application of Image Processing Method to Study the Performance of Floating Breakwater, The 19th
IAHR-APD Congress. Hanoi, Vietnam.
deVries, S., Hill, D.F., de Schipper, M.A. and Stive, M.J.F., 2011. Remote Sensing of Surf Zone Waves Using Stereo Imaging. Coastal
Engineering 58 (3), 239-250.
Erikson, Li H. and Hanson, H., 2005. A Method to Extract Wave Tank Data Using Video Imagery and Its Comparison to Conventional Data
Collection Techniques. Computers & Geosciences 31 (3), 371-384.
Foote, M. and Horn, D., 1999. Video Measurement of Swash Zone Hydrodynamics. Geomorphology 29 (1-2), 59-76.
Foti, E., Rabionet, I.C., Marini, A., Musumeci, R.E. and Sánchez-Arcilla, A., 2011. Experimental Investigations of the Bed Evolution in Wave
Flumes: Performance of 2D and 3D Optical Systems. Coastal Engineering 58 (7), 606-622.
Gonzalez, R. C. and Woods, R. E., 2002. Digital Image Processing, Prentice-Hall, New Jersey, pp. 34-51, 282-294, 567-585.
Holthuijsen, L.H., 2007. Wave in Oceanic and Coastal Waters, Cambridge University Press, New York, pp. 3-7, 24-29, 68-75, 118-131.
Hughes, S.A., 1993. Physical Models and Laboratory Techniques in Coastal Engineering, Advanced Series on Ocean Engineering 7, World
Scientific, Singapore, pp. 51-80.
Hunt, John N., 1979. Direct Solution of Wave Dispersion Equation. Journal of the Waterways, Port, Coastal and Ocean Division 105 (4), 457-
459.
Iglesias, G., Ibáñez, O., Castro, A., Rabuñal, J.R. and Dorado, J., 2009. Computer Vision Applied to Wave Flume Measurements. Ocean
Engineering 36 (14), 1073-1079.
Jain, R., Kasturi, R., and Schunck, B. G., 1995. Machine Vision, McGraw-Hill, New York, pp.1-14, 25-48, 168-181.
Kamphuis, J.W., 2000. Introduction to Coastal Engineering and Management, Advanced Series on Ocean Engineering 16, World Scientific,
Singapore, pp.54-59.
Kuo, C.A., Hwung, H.H. and Chien, C.H., 2009. Using Time-Stack Overlooking Images to Separate Incident and Reflected Waves in Wave
Flume. Wave Motion 46 (3), 189-199.
Lee, H.S. and Kwon, S.H., 2003. Wave Profile Measurement by Wavelet Transform. Ocean Engineering 30 (18), 2313-2328.
Marques, O., 2011. Practical Image and Video Processing Using MATLAB, Wiley/IEEE Press, New Jersey, pp. 335-338, 347-348, 357-358.
McAndrew, A., 2004. Introduction to Digital Image Processing with MATLAB, Thomson, Singapore, pp. 240-245.
Palmsten, M.L. and Holman, R.A., 2012. Laboratory Investigation of Dune Erosion Using Stereo Video. Coastal Engineering 60, 123-135.
Vousdoukas, M.I., Kirupakaramoorthy, T., Oumeraci, H., de la Torre, M., Wübbold, F., Wagner, B. and Schimmels, S., 2014. The Role of
Combined Laser Scanning and Video Techniques in Monitoring Wave-by-Wave Swash Zone Processes. Coastal Engineering 83, 150-165.
Wang, C. C., Chen, P.C. and Liao, C.Y., 2012. Application of CCD Cameras as a Versatile Measurement Tool for Flume Tank. Ocean
Engineering 42, 71-82.
Yao, A. and Wu, C.H., 2005. An Automated Image-Based Technique for Tracking Sequential Surface Wave Profiles. Ocean Engineering32 (2),
157-173.