INNOVATIVE TECHNIQUES OF HYPER-SPECTRAL AIRBORNE IMAGERY PROCESSING ON TEST AREA AS COMPARED WITH GROUND-BASED FOREST INVENTORY DATA IN TVER REGION, RUSSIAстатьяИсследовательская статья
Дата последнего поиска статьи во внешних источниках: 11 апреля 2019 г.
Аннотация:Remote sensing systems might be considered as an alternative to groundbased
forest inventory dealing with laborious works on test sites. Pre-requisites of
the newly defined techniques are presented to recognize remote sensing images and
to estimate parameters relating to the forest canopy of different species and age on
the test area, where common-used ground-based map of forest inventory is available.
The test site is described of the area in Tver region of Russia based on domestic
hyper-spectral imagery processing using spectral and texture features of the forest
objects extracted from the images. The recognition procedures are presented of the
airborne imagery processing with the errors of the relevant validation together with
such parameters of the forest canopy as the projective cover and biomass amount
retrieved by the related techniques for different types of forests. The accuracy
category is typical within 5-15% for the trial plots in the inventory maps. The
accuracy of our pattern recognition techniques for hyper-spectral imagery processing
is not worse of this level. This opens up prospects for the newly defined applications
of remote sensing imagery processing.
