DIGITALNA ARHIVA ŠUMARSKOG LISTA
prilagođeno pretraživanje po punom tekstu




ŠUMARSKI LIST 9-10/2008 str. 43     <-- 43 -->        PDF

D. Klobučar: PRIMJENA HISTOGRAMA DRUGOGA REDA U PROCJENI RELATIVNOG SASTOJINSKOG ... Šumarski list br. 9–10, CXXXII (2008), 419-429
Namely, instead of only one point, any pair of picture points may be observed
in a similar way, whose mutual position is determined with a relation. In
this case, the assumed random experiment becomes two-dimensional, and
evaluation or estimation of a probability density function is obtained from a
second order histogram.


This is the reason that researchers in the field of remote sensing use data
obtained from second order histograms to analyze texture and perform classifications.
First order histograms are less frequently used. Texture features obtained
with second order histograms are used in forestry remote sensing, especially
for purposes of texture classification or segmentation, but are rarely
used for stand parameter estimation.


This research indicates yet another possibility of applying digital image
analysis to the assessment of relative density and stand condition. Second order
histograms may be used to evaluate stand density. Three basic histogram
shapes were constructed which can be associated with the related density categories.


Cyclic aerial photogrammetric survey of the Republic of Croatia is an exceptionally
important project for space management. Hence, the application
of second order histograms in remote sensing for the needs of forestry and for
forest management in particular, shows continuity of research into the use of
cyclic imaging in forestry. At the same time, it is yet another contribution of
forestry to this project.


According to the available literature treating this issue, research in the
past generally involved histogram analysis for the purpose of automatically
recognizing crowns in the images during the vegetation period. Histogram
analysis, especially in pictures obtained during vegetation dormancy, i.e.
when vegetation is without leaves, is a relatively new approach.


Procedures of digital picture analysis are still imperfect and cannot fully
replace classical procedures. Consequently, integrating digital and visual
methods in decision making (with the goal of sustainable forest management)
is considered appropriate.


K e y w o rd s : cyclic records, density, first and second order histograms,
digital ortophoto, texture.