It has been observed that overall classification accuracy giving satisfactory
results, FCD mapper semi-expert system shows 80.21% accuracy followed by
object oriented image analysis of 87.50% and 71.88% respectively. The
correlation coefficient value of FCD model with visual interpretation and image
segmentation are found to be 0.95 and 0.84 respectively.
Delineation of forest vegetation from the other objects is considered to be
very important factor for precise analysis of forest change detection and landuse
processes. FCD model shows acceptable degree of delineation of forest vegetation
from the other non-forest classes. However, same inputs used for analysis in
image segmentation of eCongnition v2.1 and unsupervised cluster analysis of
Erdas Imagine v8.5 shows boarden inter mixing of forest vegetation with other
classes