January 13, 2007

Using an Underwater Camera to Monitor Real-Time Fish Activities

By Bradbeer, Robin S; Ku, Kenneth K K; Yeung, Lam F

Camera Operates With Low-Disturbance to the Environment of an Inshore Coral Reef

The most common methodology for monitoring fish activity on coral reefs is to use scuba divers. Several articles have been published in mis area.

However, any form of human intervention can be damaging to the reef, especially in areas where the habitat is under threat. This is especially the case where inexperienced divers are involved.

The work for the project discussed in this article was carried out at the Hoi Ha Wan Marine Park in Hong Kong. This is one of the few areas in Hong Kong where the habitat is nominally conserved and, being a sub-tropical area, corals grow in patches instead of the extensive formations found elsewhere. City University of Hong Kong has a laboratory attached to the World Wildlife Fund Marine Life Centre in Hoi Ha-an area used for a number of activities related to marine science and engineering. A few articles have already been published with great detail on the use of underwater instrumentation and robotics,3,4 as well as the use of remolely operated vehicles to conduct environmental monitoring on the reefs at Hoi lia Wan5.

This article describes an underwater camera and instrumentation system for monitoring fish species activity on an inshore coral reef in the Hoi Ha Wan Marine Park. The system consists of a high- resolution pan, tilt and zoom camera with associated lighting, and an instrumntation package for measuring the local environment, including dissolved oxygen, temperature, pressure, salinity and ambient light conditions. It is connected to the shore base station via a fibre optic cable with power conductors. The system is designed to be continuously used for long periods of time. The first period chosen was for 100 days during the coral spawning season from June to August 2004. The system has also been used regularly for two to three weeks at a time to monitor changes in fish behavior in different climactic conditions over the past two years.

Results from the initial experiment show that fish species activity is markedly different from that recorded by human divers. There is also some correlation between species, time of day and climatic and environmental conditions-correlations that have not been reported before. This article provides some results of the observations, as well as technical details of the system.

Camera System

The entire system uses bidirectional communication via an underwater cable with two multimode optical fibres, providing one video channel and a forward and backward digital transmission for the camera, sensors and lighting control. The optical channel is a combination of video (with a wavelength of 820 nanometers) and sensor data. With wave division multiplex filtering, the sensor data and video is recorded to a DVD and a computer, respectively. The main power provided to the underwater system is 110 volts, stepped down from an isolation transformer of 220 volts. A specially designed cable with two multi-mode optical fibres and two twisted copper power cables was constructed for this project. The cubic was coated with a PVC waterproof covering with aramid yarn to prevent water leakage along the cable. Its diameter was 14 millimeters, with a length of 210 meters.

The camera was based around a SpeedOome III unit. This has a resolution of 480 lines, with an output composite video of one volt peak-to-peak in the National Television Systems Committee format. The SpeedDome was chosen because it had the ability to track objects using built-in software.

The underwater housing for this project includes an acrylic dome (acting as a window for the camera), a main body (to house the electronics) and a scaled cover with attached underwater cables. The acrylic dome is screwed on the top of the main body with an O-ring seal. Acrylic and nylon were used as the casing construction materials. The housing contained connectors for serial data to and from the instrumentation and the intensity control of the light- emitting diode (LED) light.

Methodology and Results

The video camera was deployed on a hard coral community (around five meters in depth) at the Marine Life Centre Bay in the Hoi Ha Wan Marine Park. Additional lighting tor nighttime recording was provided by an LED light controlled via the communications system, in addition to two higher intensity lights which could be switched on automatically using a timer switch.

Ten-minute video footage for each hour was taken around the clock during the month of July 2004. The data SLM offish assemblage representing fine days were collected for five days (July 4, 7, 8, 10 and 15). Another data set representing days of storm and heavy rainfall, which occurred between July 16 and 18 (n = 3), were also obtained. The visibility of the water was approximately two meters. The camera recorded video footage of a water volume of around two cubic meters. Species were identified and their number in each footage was recorded. The number of fish was quantified as the number of fish viewed in each 10-minute video period. The density of fish was obtained hy dividing the number of fish by the volume of water (i.e., two cubic meters).

Abundance and species of fish were identified from this footage. The number offish species and levels of fish density generally increased during daylight and decreased to almost zero at night. These two parameters were also the highest during dawn and dusk. The data shows that the coral fish are most active during dawn and dusk in fine Jays. At night, they tend to hide in crevices among the coral colonies for shelter. There is also an increase in fish density in the moming of stormy days. This indicates that the fish tend to actively teed within a brief period of time.

The camera has been deployed in similar fashion since the middle of June 2005. Preliminary results show slightly different weather conditions. However, on the night on June 30, 2005, egg masses were observed being ejected from coral colonies (Faviidae). Ejection of sperm occurred subsequently. and caused the translucent nature of the water column.

Visual data from the video footage of this spawning event is currently being cross-correlated with data from the attached instrumentation package, which measures salinity, dissolved oxygen, light intensity, witter pressure and temperature.

Comparison with Scuba

The fish census data set collected through the five-day period was used to compare with data collected by the scuba diving method. The scuba diving method was conducted on a 100-meter transect line at the same coral site of the underwater surveillance camera (in July 5.

In order to obtain a comparable data set, the diver stayed in the belt transect for 10 minutes for fish observation. The data shows a decrease in the number of fish counted using the scuba diver method. This decrease is signillcant and pronounced.


Traditional diver-based fish census has to be carried out on fine, warm days. However, the camera described in this article can run continuously during fine and stormy days. The collected footage expands one's knowledge on fish activities during bad weather, such as days with storms and heavy rain ur low sea water temperatures, which happens frequently in sub-tropical coral areas (like Hong Kong).

Using underwater cameras for fish census has the advantage of increasing sampling size in terms of time, but it is restricted to the spatia] sampling area. The diver method is the reverse (i.e., temporal sampling size is restricted, hut the sampling area can be increased due to the mobility of the diver).

In this study, the camera observed fish only over a five-cubic- meter coral area, whereas the diver observed an area of more than 400 cubic meters (i.e., four segments of 20 meters long by five meters wide).

When the camera is used in a coral area with a high sediment load, the dome is subjected to be masked by the sediment and "marine snow." Marine snow are tiny lumps of fine silt and sediment stuck together by plankton or mucous secreted by marine organisms. After the camera has been deployed for a few days, a biollhn consisting of bacteria and algae also lbnns on the dome. This biofilm is the first stage of fouling and attracts a settlement of more larvae of fouling organisms. The settled sediment, silt and fouling has to be cleaned regularly by a diver using cotton cloth {so as to obtain a clear image).


It is clear from the results presented in this article that monitoring fish behavior and identifying specific fish species data is more sensitive using a fixed, permanent video camera when compared to the more usual scuba diver methods.

The permanent video camera can also provide data for bad weather conditions, as well as at night.

The results presented were for a camera observing a coral reef approximately 200 meters from the onshore laboratory. Currently, two high-resolution cameras using powered DC (20volt) fibre optie cable have been deployed at two different coral areas in Hoi Ua Wan for nuire fish species data observation, and is correlated with similarly deployed instrumentation. The system is capable of working at a distance of approximately live kilo\meters from the shore station.


This project was financed with support from the City University of 1 long Kong Strategic Research Grant Number 700141S.

The authors would like to thank Dr. Katherine Lam of City University of Hong Kong, Department of Biology and Chemistry, for her fish species data analysis and dedication to this project,


1. C'nlton. D. K, ami W. S. Levizon, "Diurnal Variability in a Kish Assemblage of a Bahamian Coral Reef," Environmental Biology of Fishi's, vol. 6, pp. .141-345, 1981.

2. Robhlee, M. B. and J. C'. Zieman, "Diel Variation in the Fish Fauna of a Tropical Scagrass Feeding Ground," Bulletin of Murine Science, vol. 34, pp. 335-345. 1984.

3. Ku, K. K., R. S. Bradbeer, K. K. Lam, L. F. Ycung and C. W. Li, "An Underwater Camera and Instrumentation System lor Monitoring the Undersea Environment." Proceeding of the 10th IEEE International Conference on Mechatronies and Machine Vision in I'racticc, pp. 1X9- 194, 2004.

4. Bradbeer, R. S., K. K. I am. L. F. Ye u 1 y and K. K. Ku, "Real-Time Monitoring of Fish Activity on an Inshore Coral Reef," Proceedings of MTS/Oceanx 2005 Conference, vol. I, pp. 542- 548,2005.

5. Lam. K. K., P. K. S. Shin. R. S. Bradheer. D. Randall, K. K. Ku. P. Hodgson and S. G. Cheimg, "A Comparison of Video and Point Inlereept Transect Methods for Monitoring Subtropical Coral Communities," Journal of Experimental Marine Biology and Ecology, vol. 333, Issue l.pp. 115-128.2006.

For more information on this subject matter, visit our Web site at www.sca-teehnology.com and click on the title of this article in the Table of Contents.

By Dr. Robin S. Bradbeer

Director/Asociate Professor

Kenneth K. K. Ku

Research Assistant


Dr. Lam F. Yeung

Associate Professor

Electronic Engineering Department

Hoi Ha Wan Marine Science and Engineering Laboratory

City University of Hong Kong

Hong Kong, China

Dr. Rohm S. Bradhecr is an associate professor of the Depurtmeni of Electronic Engineering, City University of Hong Kong, and di- rector of the Marine Science and Engineering Laboratory, Hoi Ha Wan Marine Park, leading a team working on underwater robotics and instrumentation in conjunction with marine biologists.

Kenneth K. K. Ku is a resetirch assistant in the Department of Electronic Engi-neering, City University of Hong Kong. While he is currently working on hydraulic muscles used as undenvuter actuators, Ku previously worked on the design and implementation of underwater cameras.

Dr: Lam F. Yeung is an associate pnifessor in the Department of Electronic Engineering, City Unirerxity of Hong Kong. Yeutifi'x current research interests include the development of robust multi- variuble control methods, large-scale systems, adaptive control systems, autonomous guidance vehicles and reallime svstemx.

Copyright Compass Publications, Inc. Dec 2006

(c) 2006 Sea Technology. Provided by ProQuest Information and Learning. All rights Reserved.