Image Processing Reference
In-Depth Information
Briefly, we illustrate the application of the image nearness measures for a collection of mi-
crofossil images. The coverings (with tolerance classes) for the Photographer in Fig. 12.2a
is shown in Fig. 12.2b. For simplicity, we have chosen average greyscale level and entropy
(information content) as the features used to define image coverings and measure similarities
between images. The subimages in Fig. 12.2a delineate tolerance classes (i.e., sets of subim-
ages with similar grey levels within a particular tolerance) that are subregions of the original
images in Fig. 12.2b. The tolerance classes in this image are dominated by
(light grey),
(medium grey) and
(dark grey) subimages along with many very dark
subimages in
Fig. 12.2b.
(12.2a) Photographer
(12.2b) Photographer TNS
FIGURE 12.2: Photographer Tolerance Classes
FIGURE 12.3: Sample Microfossil Images
The Meghdadi toolset for measuring nearness between images has been used to obtain
the measurements reported in Table 12.4. This toolset was reported in (Meghdadi et al.,
2009) and not described again here. Nearness measurements using the tiNM measure are
not included in the results reported in this section because the tiNM measure in (12.9) is
mathematically easy to understand but, so far, the implementation of tiNM is computa-
tionally very slow. The images in Fig. 12.3 show two different types of microfossils (Arm-
strong and Brasier, 2005), namely, Conodants in Figs. 12.3a, 12.3b 12.3c and Ostracods in
Figs. 12.3d, 12.3e, 12.3f, 12.3g, respectively. Among the comparisons of the pairs of microfos-
sil images recorded in Table 12.4, the most remarkable is the comparison between Fig. 12.3d
and Fig. 12.3g (these are microscope images of fossils from different rocks on micrometer
Search WWH ::

Custom Search