Why is the targetpro image so big, Why is the target image so “pixelly – Triton TargetPro User Manual
Page 82
June 2004 TargetPro™ User’s Manual
All images are from a Klein 3000 sonar, High Frequency channel operating at range of
100m and are from Lake Champlain, recorded by Middlebury College VT.
Why is the TargetPro image so big?
Modern side-scan sonars have higher resolution than older types, in this example the
Klein 3000 is recording 3200 samples per channel, each channel is 100m which gives a
theoretical across track resolution of 100/3200 = .03125m, which in turn means that each
pixel in the across track direction is equal to .03125m on the seabed.
The object of speed correction is to obtain an undistorted image, meaning that the
proportions of the image must be equal, therefore if we choose a square image of
256pixels x 256pixels we have to have an equivalent distance, on the seabed, along track
(the vertical dimension of the image) and across track, (the horizontal dimension of the
image).
In the along track direction the resolution is much lower and depends on the sonar ping
interval and the vessel speed, the faster the vessel travels the fewer pings we get per
meter. Typical along track resolutions are around 0.25m
To get equal proportions we need 256 pings to keep the image proportions square:
To obtain a sonar range of 100m the ping interval is 133.3mS (0.1333secs).
256 pings at 0.1333secs is 34seconds of data in the along track direction.
The vessel speed was around 3.26Kts or 1.76meters/sec.
So in the along track direction we have 34x1.76 (about 60m) of data.
In the across track direction we have .03125m per pixel, therefore the image ends up
being 60m x 60m or 1920 pixels square.
In the example we have 1922x1922
Why is the Target image so “pixelly”
Target solves the problem the other way. It maintains the image proportions by fixing the
number of across track pixels at 256 and then figuring out what is required in the along
track direction to keep the image proportions correct, so we end up with an image that is
256x256 pixels at the across track resolution:
256/0.030125 = 8m
If we were to extract the centre 256x256 pixel (8m x 8m) “box” from the middle of the
TargetPro image it would look exactly like the Target image:
Addendum: The Effect of Speed Correction on Images from TargetPro and Target
71