The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences. Vol. XXXVII. Part B6b. Beijing 2008 
Figure 3. Fork lengths of grass carp(upper) and common 
carp(lower), respectively. 
The real fork length of grass carp and common carp is about 
6-8cm and 10-12cm, respectively. The results indicated that the 
uncertainty in the measurement of the fork length is 1 cm. We 
believe that it is possible to improve the measurement accuracy. 
As the 3-D sample block is reconstructed in units of mm, the 
uncertainty in fork length is due to arithmetic error and not the 
DLT process itself. First, it is important to choose the control 
points on the fish carefully, e.g., the same point on the head or 
tail at the same time. Second, fish are not necessarily straight 
objects; instead of simply taking the straight line from the head 
to the tail as the fork length, the arithmetic must be different for 
calculating the fork length of a straight or twisted body. 
4.2. Swimming Speed 
The digital hard disk video recorder used in this experiment can 
record only nine pictures in 2 s. To determine the swimming 
speed, the positions of the fish, as indicated by certain features 
on the body, are obtained from the different pictures. Figure 4 
shows six continuous pictures taken with the camera on the 
right side. 
Figure 4. Six pictures recorded with the right camera. The white 
triangle marks the same fish. 
By 3-D DLT reconstruction of the world coordinates of a 
certain feature in the fish using multiple pictures from both 
cameras, the swimming speed of fish for five different time 
intervals were calculated as 23.2, 24.6, 22.4, 25.6, and 26.1 
cm/s. 
4.3. Orientation 
The coordinates of the camera in the 3D object space can be 
retrieved using the DLT parameters: 
*0 
A 
A 
V 
-1 
'-A' 
^0 
= 
V 
A 
V 
-A 
z 0_ 
_A 
Ao 
11 _ 
l 
Therefore, the orientations of two cameras and the relative 
position of the fish with respect to the camera can be 
determined using the camera’s DLT parameters. The orientation 
in object space of the cameras and the ten randomly chosen fish 
are shown in Fig. 5. 
Figure 5. The orientations of cameras and fish in object space. 
Blue triangles indicate the camera and blue stars represent the 
fish. 
5. CONCLUTIONS 
Digital photogrammetry can withstand longer periods under 
water than regular underwater photogrammetry. A 200 GB hard 
disk can record continuous images for seven days and nights, 
and these images can then be used to study the habits of animals 
living around an artificial reef. This paper discussed the 
feasibility of applying digital photogrammetry to monitoring the 
animals living around an artificial reef. Parameters such as the 
fork length of the fish and the swimming speed as well as the 
position of the fish can be retrieved by measuring sample fish in 
a water tank. There is scope for improvement in the accuracy of 
calibration and feature-matching. Preliminary tests indicate that 
using digital photogrammetry to monitor aquatic life around the 
reef is a feasible means of evaluating fish behavior. 
REFERENCES 
Abdel-Aziz, Y.I., Karara, H.M., 1971. Direct linear 
transformation from comparator coordinates into object space 
coordinates in close-range photogrammetry. Proceedings of the