High Spatial Resolution Digital Imagery
53
conditions during image acquisition meant that the deciduous trees and shrubs
were without leaves. In addition, the reed and grass species were fully senesced.
The lack of photosynthetically active vegetation may have somewhat diminished
the overall interpretability of the NDVI image by reducing the contrast between
vegetated and non-vegetated segments. However, the NDVI image still was able
to adequately separate those pixels that were primarily vegetation from those
pixels that did not include vegetation. This is based on near infrared reflectance
properties of chlorotic and necrotic vegetation being still very different from the
near infrared spectral response of non-vegetated features, such as water and bare
soil.
NDVI Histogram
Normalized Difference Vegetation Index =
(band 4 - band3) / (band4 + band3)
or
(NIR - Red) / (NIR + Red)
Vegetation Mask
Non-vegetation Mask
Figure 31. NDVI image (left), NDVI histogram (upper right), and primary
NVDI masks.
The segmented images for each mosaic were converted into two masks. An
image mask assigns a value of 1 to pixels that are "switched on" and a value of 0
to pixels that are "switched off" (Fig. 31). The non-vegetation mask had all non-
vegetation pixels assigned a value of 1, and all vegetation pixels assigned a value
of 0. The vegetation mask maintained the reverse values (i.e., veg = 0, non-veg =
1). Each mask was independently applied to the original four-band mosaic creat-
ing two unique multispectral images (Fig. 32). One image contained primarily
vegetated pixels and the other contained the remaining non-vegetated pixels.
Previous Page
