The sonar survey was conducted during daytime in 
December 1995 at Funka Bay using the R/V Ushio- 
Maru, which is equipped with acoustic instruments. 
The sonar used was a 180-degree sector scanning 
sonar model KCH-1827 (Kaijo Co. LTD.), with a 
receiving beam width of 8 degrees using the 
acoustical frequency of 164 kHz. Sonar images were 
displayed in 16 colors according to the echo levels. 
The sonar beam was directed downward and 
perpendicular to the ship's course to obtain vertical 
sectional images while the ship was cruising (Figure 
1). 
    
  
Bottom : 
Figure 1 Cross sectional image sampling by 
the vertical use of sector scanning sonar beam. 
Analog RGB video signals of the sonar images were 
converted to NTSC video signals through a video scan 
converter and recorded on videocassette. The video 
signal was then replayed and digitized once every 1-2 
seconds with a resolution of 256 by 256 pixels with 18 
bits using a video capture interface to a personal 
computer. Ordinary echograms were also recorded 
using a vertical scientific echo-sounder (model 
KFC200, Kaijo Co. LTD). 
2.2 Three-Dimensional Analysis of Sonar Image 
Since digitized sonar images contain noises and color 
distortion, a median spatial filter was used to eliminate 
noise and to emphasize the edges of the images. The 
data value was also reduced from 262,144 (64 x 64 x 
64) to 256 to hasten the three-dimensional processing. 
About 600 pre-processed sonar images were stored 
as one volume file and then analyzed using the three- 
dimensional processing software SLICER (Fortner 
Research LLC). 
Volume data was composed of stacked sectional 
images sampled every 1.5 seconds (Figure 2). The 
thickness of volume data D is expressed by the 
product of the sampling period, ship speed, and 
number of images. However it was still impossible to 
observe fish schools or the sea bottom because of the 
Ship 
  
    
  
m 
X 1.5 séc 
Figure 2 Continuous image sampling 
and volume data file of sonar image. 
A Obstacle area 
  
Figure 3 Extraction and reconstruction of 
three-dimensional fish school image by sonar 
image processing. 
730 
Of 
Wi 
th 
rc 
th 
th