NCEAS Project 5081

Evaluation functions for ecological image segmentation

  • William T. Langford

ActivityDatesFurther Information
Postdoctoral Fellow1st February 2002—31st January 2005Participant List  

Abstract
This project examines and enhances various forms of evaluation functions used in the comparison and optimization of image segmentation algorithms. It defines quantitative criteria for definition of the evaluation functions and uses them in the design and selection of these functions. It also investigates the relationship between errors in landscape pattern metrics and segmentation and classification algorithms. Finally, it coordinates the creation of a repository of correct segmentations of aerial and remotely sensed ecological imagery to improve algorithm development by making rigorous comparison of algorithms possible.

TypeProducts of NCEAS Research
Journal Article Langford, William T.; Gergel, Sarah E.; Dietterich, Thomas G.; Cohen, Warren B. 2006. Map misclassification can cause large errors in landscape pattern indices: Examples from habitat fragmentation. Ecosystems. Vol: 9. Pages 474-488.
Journal Article Moret, Stephanie L.; Langford, William T.; Margineantu, Dragos. 2006. Learning to predict channel stability using biogeomorphic features. Ecological Modelling. Vol: 191. Pages 47–57.
Journal Article Regan, Helen M.; Ben-Haim, Yakov; Langford, William T.; Wilson, Will G.; Lundberg, Per; Andelman, Sandy J.; Burgman, Mark. 2005. Robust decision-making under severe uncertainty for conservation management. Ecological Applications. Vol: 15(4). Pages 1471-1477.
Journal Article Wilson, Will G.; Lundberg, Per; Vazquez, Diego P.; Shurin, Jonathan B.; Smith, Melinda D.; Langford, William T.; Gross, Katherine L.; Mittelbach, Gary G. 2003. Biodiversity and species interactions: Extending Lotka-Volterra community theory. Ecology Letters. Vol: 6. Pages 944-952.