Table 1.
Flight details for multiple flight altitudes.
Fig 1.
Proposed algorithm.
Fig 2.
Workflow of proposed method.
Table 2.
Orthomosaic processing details for altitude variants.
Fig 3.
(a) Color histogram of orthomosaic (b) Color histogram of orthomosaic after LCS.
Fig 4.
Study area separated into training region and ROI (top right).
Table 3.
Training details of altitude variants.
Fig 5.
Sample image tiles of 40m altitude variant.
(a) LCS (p1), (b) SCT (p2) and (c) TGI (p3).
Fig 6.
Detection results on variants of image processing methods on a ROI tile of 40m.
(a) LCS (p1), (b) SCT (p2) and (c) TGI (p3).
Fig 7.
Merging detection results on a sample ROI tile of 40m.
(a) Bounding boxes from LCS (Black), SCT (White) and TGI (Red), and their centroids (small blue circles) on a sample image tile of ROI. (b) Merged centroids by taking euclidean distance threshold of 30 pixels are represented by blue circles of 15px radius.
Fig 8.
Final detection on ROI of 40m altitude variant.
Fig 9.
Final detection on ROI of 50m altitude variant.
Fig 10.
Final detection on ROI of 60m altitude variant.
Table 4.
Detection performance on ROI of 40m altitude variant.
Table 5.
Detection performance on ROI of 50m altitude variant.
Table 6.
Detection performance on ROI of 60m altitude variant.
Table 7.
The transform parameters.
Fig 11.
Detection results of combination (40+50)m.
Fig 12.
Detection results of combination (40+50+60)m.
Table 8.
Detection performance after combination of altitude variants.