Suraj AmatyaWashington State University/Biological Systems Engineering
: + 1 5097885043 Email: [email protected]
GoalDetecting cherry tree branches and locating shaking positions for automatic harvesting
Conclusion• Automated cherry harvesting is possible using machine vision for branch detection
• Morphology of branch sections and cherries can be used to detect branches in full foliage canopies
• Overall 91% of branches detected
• 93.5% cherries harvested by shaking at points picked by the algorithm
Branch Detection
• Estimated branch segment featureso Orientation, Major Axis Length, Minor Axis Length
• Grouped segments of the same branch
• Merged branch sections using region growing
algorithm
• Fitted linear/logarithmic equations to branches
Cherry Detection
• Identified and segmented cherry clusters
• Grouped cherry clusters located along a
vertical direction
• Determined orientation of branch • Fitted branch equations
Shaking Point Localization
• Located cherries in detected
branches
• Divided tree canopy into 3 shaking
zones
• Shaking position identified close to
larger cherry clusters
• Obtained distance of shaking points
from 3D camera
Original Image
Segmented
Branches
Grouped Branch
Sections
Detected Branches Original Image
Vertical Cherry
Cluster Search
Detected Branches
Localization of shaking points for harvesting
Results
HarvestedNot harvested (within FOV)
Not Harvested (beyond FOV)
FOV
= Field O
F View
Weight (lb.) 306.7 12.0 9.4
Percent (%) 93.5 3.7 2.9
Table 2: Results of cherry harvesting trials
Table 1: Results of branch detection method
based on branch pixels and cherry pixels
Actual Detected
False
Detection
True
Detection Undetected
Vert
ical
No. of Branches 453 477 73 404 49
Percentage (%) 100.0% 105.3% 16.1% 89.2% 10.8%
Y-t
rell
is No. of Branches 453 481 56 425 28
Percentage (%) 100.0% 106.2% 12.4% 93.8% 6.2%