SR SE NE
AS AS À
ERE TUNER UR
Figure 4: Siemens star with 32 wedges: For smaller radii
the spatial frequency of concentric circles is increasing.
The smallest radius where all wedges are resolved char-
acterizes the quality of the image.
of the test images,
G—1
H,=— M P(k)log2[P(k)]
k
iz
for an image with G gray-levels and the probability of
gray-level k is P(k). For an ideal homogenous patch con-
taining only one gray-level the entropy is 0. Smaller en-
tropy indicates less information content and is a measure
for image noise, if calculated in a homogenous patch.
3 RESULTS
All testing methods were performed on all film images
scanned with 5, 10, 15 and 20pm and on the ccd images
with 9m which covers the same area as the 20m scans.
If we have an ideal film image and an ideal scanner we
could scan the image with e.g. double resolution and will
get the same results for e.g. the Siemens star test. As we
are dealing with real film images it is obvious that the geo-
metric performance will not stay the same when scanning
with very high resolution, because the granularity of the
film causes image noise.
In the following tables the values in brackets are the ra-
tios to the 51m scans. If the ratio is higher than the ratio
between the resolutions (i.e. between Hum and 15m we
have ratio of 3), this indicates that there is no more infor-
mation in the 5pm image than in the lower resolution one.
So if the ratio is about 2.0 for the 10m, 3.0 for the 15um
and 4.0 for the 20pm scan, then we know that the 20pm
scan contains same geometric accuracy as the Sum and
therefore scanning with 20m makes sense without losing
information.
For ideal images the ratio is 1.0 independent of the resolu-
tion.
3.1 Image matching
For the image matching setup we used stereo images with
a baseline of about 10 meters. One of the stereo images
is shown in figure 5. Then 10000 points of interest were
extracted from the reference image and reallocated in the
search image. Reference and search window size were
adapted according to the scan size. For example if we have
a search window size of 600 x 80 pixel in the Sum case,
then the window size for the 20pm case is smaller by a
International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Vol XXXV, Part B3. Istanbul 2004
factor of 4, so 150 x 20 pixels. This guarantees that these
windows covers the same area in the real object.
VE
Figure 5: One image of the stereo scene taken with ccd
sensor. Shown are 1800 x 1200 pixels.
Figure 6 shows the histograms of the absolute distances to
the epipolar lines for film delta scanned with 5, 10, 15 and
20um. The histograms are scaled so that each histogram
shows the same distances in object space (e.g. 8 pixels in
Spm are 2 pixels in a 20pm scan). As quality measure we
choose full width at half maximum, called c. The ratios of
these c values are 2.06, 2.91 and 3.12 which means that
the 15m scan leads to same accuracy as the 5um scan,
but scanning with 20pm leads to a slight loss of quality.
All results for this test are given in table 1.
Figure 7 shows the histograms for 20m scans and for the
ccd image. The number of matched points is significantly
higher for ccd case and the o value of ccd outperforms the
film with a factor of 2 to 3, which means on average 2.5
times better image matching results.
Table 2 shows the maximal absolute distance for the best
1000 matches and here the accuracy in 20pm scan is even
better than using the 5m scan. The digital ccd image out-
performs the 20m scans by a factor of 3.0 to 5.0.
1138
oum 10um 151m 20pm
apx 2.078 1.156 0s) | 0.724 ox» | 0.656 545
delta | 2.000 | 0.969 coe) | 0.688 299 | 0.641 63.12)
t-max 1.950 0.924 (2.11) 0.691 (282 0.453 (4.30)
t-pan 1.989 1.042 non | 0.732 em | 0.453 (439)
ced Yum - - - 0.219
Table 1: Full width at half height of distances to the epipo-
lar line in pixels (values in the brackets are the ratios to
oum scan).
oum 10pm 15um 201m
apx 1.232 0.597 (2.06) 0.386 (3.19) 0.272 «s»
delta | 1.080 | 0.397 2», | 0.288 «s» | 0.229 «m
t-max 0.739 0.278 (2.66) 0.226 (3.14) 0.183 (4.04)
t-pan 0.803 | 0.315 (ss, | 0.268 ov» | 0.171 am)
ced 91m - - - 0.0557
Table 2: Maximal distances to the epipolar of the 1000
best matches in pixels (values in the brackets are the ratios
to Sum scan).
In
p