Download (PDF 135 KB): https://research.fs.usda.gov/download/treesearch/3199.pdf
Abstract
Automated image analysis for forestry applications is becoming increasingly important with the rapid evolution of satellite and land-based remote imaging industries. Features derived from line information play a very important role in analyses of such images. Many edge and line detection algorithms have been proposed but few, if any, comprehensive studies exist that evaluate performance in a scientifically meaningful way. In this paper, we introduce an objective evaluation paradigm. We also demonstrate, using this paradigm, improved performance on edge and line detection. We reduced the detection error rate from 42 percent to 29 percent for 159 manually labeled forest images.
Citation
Long, Zhiling; Picone, Joseph; Rudis, Victor A. 2000. The optimization of edge and line detectors for forest image analysis. In Callaos, N.; Lombardo, P.; Huber, R., eds. Proceedings: Image, acoustic, speech, and signal processing: part I of the 4th world multiconference on systemics, cybernetics, and informatics. Orlando, FL: International Institute of Informatics and Systemics: 171-176.
