5. CONCLUSION
This research has shown that when merging CIR high-spatial
resolution aerial imagery with medium-spatial resolution satellite
imagery, an object-based classification will outperform both
supervised and unsupervised pixel-based methods. This supports
the findings of similar research that has been conducted in the last
few years. In general, the addition of the leaf-on data to the
image dataset improved overall classification accuracy and it
tended to improve the user’s accuracy for feature classes such as
deciduous and grassland.