Automatic Leaf Segmentation for Estimating Leaf Area and Leaf Inclination Angle in 3D Plant Images.

Area of Science:

• Plant Science
• Computer Vision
• Agricultural Technology

Background:

• Accurate plant monitoring is crucial for effective crop management and breeding.
• Three-dimensional (3D) plant modeling provides essential spatial and structural data.
• Automated analysis of plant structures, particularly leaf parameters, remains a challenge.

Purpose of the Study:

• To develop an automated method for segmenting 3D plant leaves.
• To accurately retrieve leaf area and inclination angles from segmented 3D leaf models.
• To enhance efficiency in plant breeding and growth management through precise structural analysis.

Main Methods:

• Initial segmentation of top-view leaf images using distance transform and watershed algorithm.
• Generation of seed regions from downscaled images and re-projection onto 3D models.
• Leaf segmentation expansion using 3D information and voxel-based calculations for parameter estimation.

Main Results:

• Successful automatic segmentation of individual leaves from 3D plant models.
• Accurate estimation of leaf area and leaf inclination angles post-segmentation.
• Demonstrated efficiency and accuracy in plant structure analysis.

Conclusions:

• The developed automated 3D leaf segmentation method enables precise plant structure analysis.
• This technique facilitates accurate and efficient plant breeding and growth management.
• The voxel-based calculation provides reliable estimations of leaf area and inclination.