netric parameter 
c,-map), specula 
coefficient (c,- 
2. 9, respectively. 
er-left and lower- 
. 6, respectively; 
7, 3 and 0 in Fig. 
ows the synthetic 
del and c,-map, 
d c,, c, and c,- 
uth and 45? slant 
rt height image, 
ge by using the 
ing the Torrance- 
direction of light 
g. 29 and Fig. 34. 
IM, we down- 
vs and every four 
ints between the 
precision height 
rder to show the 
| Fig. 26 by using 
> (see Fig. 10 and 
in this paper is 
> plot the middle 
Fig. 28) to show 
parts are plotted, 
ic model in every 
and Fig. 29, the 
. 34. Fig. 13--14 
after de-shading. 
g the same light 
d DTM. Fig. 17 
e light source to 
, €, and c, maps 
5 show the same 
-Sparrow model. 
sults as Fig. 13-- 
ge. It should be 
early stretched in 
) and Tab. 2 (for 
re errors between 
ing maps of the 
s between the re- 
ithetic brightness 
1eans the number 
3 to generate the 
square errors of 
38, 
orks in what we 
g the estimated 
ameters and the 
iginal brightness 
specially in the 
t inside and near 
able, and outside 
all to allow a 
to improve both 
operties and SFS 
This will be our 
  
  
  
  
  
  
  
  
  
  
  
Mars image Mozart image Mars image Mozart image 
Sn ic , map Prightness| c,-map brightness Sn ic ,map i c,-map i c,-map brightness| c, -map 1 c,-map i c,-map brightness 
2 10.0025 10.0377 | 0.0025 i 0.0024 2 10.0121 1 0.0997 1 6,4647 10.0509 | 0.0451 1 0.0319 1 6.0216 1 0.0318 
3 10.0043 § 0.0418 | 0.0022 1 0.0039 3 10.0149 3 0.0329 16.3290 10.0550 | 0.0398 1 0.0581 1 6.0016 1 0.0335 
4 10.0105 10.0448 | 0.0016 1 0.0055 4 10.0134 1 0.0563 16.2307 1 0.0489 | 0.0432 1 0.0991 16.0239 1 0.0338 
  
  
  
  
  
  
  
  
  
  
  
  
  
  
Tab.1 Mean square errors for 
the Lambertian Model. 
Fig.4 
Fig. 8 
Fig. 12 
Fig. 16 
Tab. 2 Mean square errors for the Torrance-Sparrow Model 
Fig. 9 
Fig. 17 
1033 
Fig. 14 
International Archives of Photogrammetry and Remote Sensing. Vol. XXXI, Part B3. Vienna 1996