Skip to main content
ARS Home » Northeast Area » Kearneysville, West Virginia » Appalachian Fruit Research Laboratory » Innovative Fruit Production, Improvement, and Protection » Research » Publications at this Location » Publication #337830
Title: Automatic segmentation of trees in dynamic outdoor environments

Author
item Tabb, Amy
item MEDEIROS, HENRY - Marquette University

Submitted to: Computers in Industry
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 3/2/2018
Publication Date: 3/12/2018
Citation: Tabb, A., Medeiros, H. 2018. Automatic segmentation of trees in dynamic outdoor environments. Computers in Industry. 98:90-99. https://doi.org/10.1016/j.compind.2018.03.002.
DOI: https://doi.org/10.1016/j.compind.2018.03.002

Interpretive Summary: A key part of many automated computer vision systems is a step called segmentation. In this step, image regions of interest are separated from non-interest regions by a program. Segmentation is particularly challenging outdoors because of changing illumination conditions (shadow and sunlight, etc.). We describe our work on developing computer programs to automatically segment images in challenging conditions for orchard automation applications. This work has applicability to automation tasks outdoors, such as tree fruit pruning and phenotyping.

Technical Abstract: Segmentation in dynamic outdoor environments can be difficult when the illumination levels and other aspects of the scene cannot be controlled. Specifically in agricultural contexts, a background material is often used to shield a camera's field of view from other rows of crops. In this paper, we describe a method that uses superpixels to determine low texture regions of the image that correspond to the background material, and then show how this information can be integrated with the color distribution of the image to compute optimal segmentation parameters to segment objects of interest. Quantitative and qualitative experiments demonstrate the suitability of this approach for dynamic outdoor environments, specifically, for tree reconstruction and apple flower detection applications.