Summarized "image.tbl” 
  
  
  
  
  
  
  
  
  
    
  
  
Image ID | Footprint |Pointer to lllumination &| Image 
Time etc {information| "image2.tbl"| Viewing info. | Descriptor 
"image2.tbl" 
Pointer to No. of points | Pointer to No. of points 
"footprint.tbI" "camera _pos.tbl” 
  
  
  
  
  
  
"footprint.tbl” 
*eamera pos.tb)" 
JD Imager XYZ 
JD Imager XYZ 
JD Imager XYZ 
JD Imager XYZ 
JD Imager XYZ 
JD Imager XYZ 
  
  
  
  
Y 
Det x x4 x x XI 
SEE ELS en CEST 
Ni NI NI NI NE NE 
RRARKR 
9 9,0.9 9 9 
" " un 
" " n 
  
  
  
  
Figure 2: File structure needed to represent distorted 
footprints and moving camera situations. The last field in 
"camera, pos.tbl" is used only for rotating scanner cameras. 
For the image footprint, minimum and maximum ground 
pixel size (m) are used to determine if the image is of 
sufficient resolution. The actual image footprint shape is 
stored in a three level hierarchy (fig 1) to facilitate rapid 
search. The planet's surface is divided up into NxM tiles in 
longitude and latitude directions (where N«256 and M«256, 
i.e. byte precision). N and M are chosen to optimize 
footprint storage and access. The minimum and maximum 
longitude and latitude tiles, stored as bytes, represent the 
first level of this hierarchy. To find out which images cover a 
certain area, SOLIS generates initially the corresponding tile 
IDs, then with the aid of an image-tile geographical index 
(not described here), it extracts lists of images contained in 
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
Bits Description Value 
1 Image contains the whole planet? O0 or 1 
2 Image contains a planetary limb? Oorl 
3 Image used for control point nets? Oorl 
4 Image used for topographic maps? Oor1 
S Image used for geological maps? Oorl 
6 Image referenced in publication? Oorl 
7 Image entirely blank? Oorl 
8 Image contains saturated pixels? Oorl 
9 Lossy compression used on image? Oor 1 
10-12 | Contrast: unknown 000 
10-12 | Contrast: excellent (255«40py) 001 
10-12 | Contrast: good (150«40pyX255) 010 
10-12 | Contrast: ok (80«40py150) 011 
10-12 | Contrast: moderate (40<4opn<80) 100 
10-12 | Contrast: poor (20«40py 40) 101 
10-12 | Contrast: bad (10«4o0py€20) 110 
10-12 | Contrast: terrible (4op€10) 111 
13-15 | Noise: unknown 000 
13-15 | Noise: excellent (256«S/N) 001 
13-15 | Noise: good (128<S/N<256) 010 
13-15 | Noise: ok (50<S/N<128) 011 
13-15 | Noise: moderate (12<S/N<24) 100 
13-15 | Noise: poor (6<S/N<12) 101 
13-15 | Noise: bad (3<S/N<6) 110 
13-15 | Noise: terrible (S/N<3) 111 
16-17 | Sharpness: unknown 00 
16-17 | Sharpness: good (<1.5 pixels) 01 
16-17 | Sharpness: moderate (1.5-3 pixels) 10 
16-17 | Sharpness: poor (>3 pixels) 11 
18-32 | To be decided 
  
  
  
  
  
Table 4: The image descriptor field. 
190 
each tile. This results in a reduced number of image record, 
that need to be searched. The second hierarchy level for the 
image footprint description, consists of the four corners of 4 
bounding quadrilateral which contains the footprint. The 
positions of the corners are stored relative to the minimum 
and maximum longitude and latitude tiles, to unsigned short 
integer (0-65535) precision. This is sufficient to describe 
the majority of image footprints. 
If an image is not completely covered by the level ? 
footprint hierarchy, because the footprint is distorted (fig | 
lower), then normalized two byte offsets are used to describe 
a more detailed footprint. These offset points are stored in a 
separate file "footprint.tbl" (see fig 2). 
The final field in "image.tbl" is an image descriptor, 
consisting of a 4 byte integer (see table 4). Here, bits can 
signify a positive answer (bit=1) as to whether the image 
contains some attribute of the planet, or has been used in a 
cartographic process, or in a publication. A negative answer 
(bit=0) can indicate either a "no^", or an "unknown". In 
addition, groups of bits are used to describe the image 
contrast, noise and sharpness. Additional bits are available 
for future use, for example whether an image contains a 
duststorm etc. 
     
À $ Range 
Figure 3: Top left: framing camera geometry, top right: 
push broom camera geometry, lower: rotating scanner 
camera geometry. 
Lastly, for stereo and photometric analyses, it is necessary 
to know the location of the camera with respect to the 
planetary surface. For a framing camera (fig 3 top left) 
"image.tbl" gives this, but for a pushbroom camera (fig 3 
top right), or for a rotating point source scanner camera (fig 
3 lower), the camera is moving during imaging. In figure 2 
we present a simple file structure which can be used to 
describe these, "camera pos.tbl'. Only positions and 
orientations are stored when there has been significant 
change in position and orientation. To obtain position and 
orientation in between these points, interpolation is used. 
Planet centred XYZ coordinates are used for these moving 
camera systems because they are easier to handle in 
geometric calculations involving motion. 
International Archives of Photogrammetry and Remote Sensing. Vol. XXXI, Part B4. Vienna 1996 
are 
par 
res 
im: 
dat 
nut 
for 
col 
(Jo 
wa 
imé 
pix 
Ba 
dat 
mu 
mu