Geography Reference
In-Depth Information
Upstream
Downstream
250
200
Gravel
bar
150
Transect
Water
channel
100
50
0
500
1000
1500
2000
2500
Gravel bar
N
Location
Water channel
W
E
Image 1
Image 3
Image 2
S
0
25
50 Meters
(a)
(b)
Figure 8.9
Radiometric variability in 3 consecutive images, Ste-Marguerite River, Quebec, Canada. a) Georeferenced images with
one image noticeably brighter. b) Radiometric profile for the red band along the transect in a). The water in Image 3 is clearly
brighter than in Image 2 without any significant change in real depth.
Upstream
Downstream
250
Gravel bar
200
150
Transect
Water
channel
100
50
0
500
1000
1500
2000
2500
Gravel bar
Location
N
Water channel
Image 1
Image 3
W
E
Image 2
S
0
25
50 Meters
(a)
(b)
Figure 8.10
Radiometric normalisation procedure applied to the data in Figure 8.9, Ste-Marguerite River, Quebec, Canada.
a) Georeferenced imagery. b) Radiometric profile along the same transect showing improved correspondence between the brightness
levels in all 3 images.
is low and larger portions of the channel are affected
by shading. Consequently, any attempts to quantify the
depth of a channel section based on its radiometry using
established methods (e.g., Winterbottom and Gilvear,
1997; Conyers and Fonstad, 2005; Carbonneau et al.,
2006) will result in a sharp depth discontinuity since
the darkened, shaded, areas will yield falsely exaggerated
depths. Furthermore, higher-level analyses such as the
extraction of particle sizes in rivers (Carbonneau et al.,
2004), stream hydraulics (Lorang et al., 2005), and stream
habitats (Wright et al., 2000; Marcus et al., 2003; Legleiter
and Goodchild, 2005; Hedger et al., 2006) would be as
much or more influenced by the effects of shadows as
is depth, so a basic approach to dealing with shadows is
crucial in river studies. Therefore, a transferable method
designed to remove the effects of shadows would make
the widespread application of image based bathymetry far
simpler and more accurate.
Once again there are no methods published in the lit-
erature which could alleviate the shadow problem. Given
the effectiveness of histogram equalisation in normalising
brightness values for entire images, it was also investigated
Search WWH ::
Custom Search