3. Examples of olive trees in imagery
used
Figures 5 to 10 illustrate some of the conditions
encountered in the testing carried out with
Olicount. The typical problems shown
(shadowing, irregular stands, etc.) still force the
intervention of an operator for visual checking
of the count algorithm results.
4. Trials and operational use
4.1 The Olistat project
In mid-1997, the European Commission
launched a project for the “Estimate of the
number of olive-trees by Member State" (EC
1997). The main goal of this project - dubbed
Olistat - was the production, for the five
Member States concerned, of national-level
estimates for the number of olive trees, on the
? Olistat - Check
{ JIRCCHECK
of PLACETTES
L1
of 1dm204 tif
basis of a statistical sample. The data source
for this operation was aerial photography with
characteristics as described above (§1.3).
Semi-automatic counting methods were
applied in four (out of five) sub-projects, but
were not imposed as part of technical
specification. Indeed, for small numbers of
trees in very mixed environments (a randomly
located sample plot may fall in very mixed
zones) manual counting by well-trained
interpreters can be a fast and efficient method.
However, GIS management of the entire
database was stipulated, and per-tree
identification required as a part of quality
assurance procedures.
Technical management of the Olistat projects
was conducted by the JRC, as well as a
number of aspects of quality assurance: these
included checking of photointerpretation
results.
The data set at the end of the project (April
Select IDM | Select image]
Placette Navigation
Auto Count Edit Accept / Reject
# Al 8
*
2| * e
PLACETTE INFORMATION—
i | Placette Number 97
Contractor Count (In/Out): 54 / 86
Current Check Total: 143
Of Which:
Copied from Contractor (In/Out):
Deleted from Contractor(In/Out):
Operator Added (In/Out): 0 / 0
Operator Deleted (In/Out): 0 / 0
OTCount Added (In/Out): b2 / 91
OT Count Deleted (In/Out): 0 / 0
Figure 11: The Olistat photointerpretation tool interface
362 International Archives of Photogrammetry and Remote Sensing. Vol. XXXII, Part 7, Budapest, 1998
fu.
PS nme un d em—
im (A 0 = NN =