Recognizing Plant Tension in Plantations by use of UAVs Visible Light Detector. (Case Study: Nekazalemrood Forestry Plan)

The use of lightweight and cheap UAVs to detect the health of forests and identify the tension of planted can be useful to prevent the spread of pests and diseases. In the present research, a lightweight quadcopter drone with a 12-megapixel camera, visible light range was used. This UAV was emploied to detect leaf tension of pure Quercus Castanifolia plantation, pure Acer Velutinum and their mixture in nine sample with 1-3 hectares. Flight at altitudes of 40, 70 and 100 meters was used to determine UAV ability for detecting areas plantations tension. The flight plan was designed in the form of 75% latitude coverage and 80% longitudinal coverage. Supervised classification such as Neural Net, Support Vector Machine (SVM), Maximum Likelihood and Mahalanobis Distance algorithms are used and 25% of samples were used to check the classification accuracy. Visible color saturation image and some vegetation indices such as vegetation index (NGRDI) and (EXG), has great potential for detecting leaf tension in trees and seedlings. The Jeffries-Matusita coefficient ranged from 1.81 to 1.97, and the Transformed Divergence was 1/87 to 1.98, indicating the degree of separation of educational samples. The overall accuracy of Support Vector Machine (SVM) algorithm as best method was 83 to 96.7 percent for all samples and the kappa coefficient was 0.89 to 0.98. The results revealed the high capability of visibility light sensor cameras mounted on a UAV in detecting tree leaf tension. The best flight height is between 70-100 M. Using image enhancement techniques, especially color saturation and vegetation indices, the range of visible light spectrum such as vegetation index (NGRDI) and (EXG) to detect leaf tension increase the effectiveness of these images. Design of an automatic imaging system adapted to the altitude variation of the tree crown is recommender in order to prevent a minimum level of overlapping.