Skip to main content
U.S. flag

An official website of the United States government

Finding a good segmentation strategy for tree crown transparency estimation

Informally Refereed

Abstract

Image segmentation is a general term for delineating image areas into informational categories. A wide variety of general techniques exist depending on application and the image data specifications. Specialized algorithms, utilizing components of several techniques, usually are needed to meet the rigors for a specific application. This paper considers automated color image segmentation of foliage representing pixels for the purpose of crown transparency estimation. Varying image characteristics caused by differing levels of foliation and fluctuating lighting conditions present a unique challenge for consistent segmentation in order to reliably measure tree crown transparency. Small leaves surrounded by bright sky, small openings among dense canopy, and mixed pixels all present difficulties for segmentation. This study found the green band to be most helpful for preserving pixels representing reflectent leaves surrounded by pixels representing sky. Texture was useful for preserving other small portions of tree canopy in sky dominated regions. Scale considerations and automated methods are also discussed. The initial stages of a segmentation approach are presented, but further work is required for segmenting pixels in the fuzzy regions.

Citation

Clark, Neil A.; Lee, Sang-Mook; Araman, Philip A. 2003. Finding a good segmentation strategy for tree crown transparency estimation. In: Proceedings of the 19th Biennial workshop on color photography, videography, and airborne imaging for resource assessment. Finding a good segmentation strategy for tree crown transparency estimation 1-57083-074-6. Logan, UT: American Society for Photgrammetry and Remote Sensing: 10.
https://www.fs.usda.gov/research/treesearch/7660