Responding to an increasing interest in studying vegetation changes over time, we review current methods of processing black and white digital aerial photographs in order to classify tree cover in pinyon-juniper woodlands. Besides applying commonly used clustering and supervised maximum-likelihood methods, we have developed a new classifier, nearest edge thresholding, which is unsupervised and based on the principals of edge detection and density slicing. Comparison of the three methods' abilities to predict field values at plot scales of 100 m2 to 900 m2 shows this new method is better or comparable to others at all scales, can be easily applied to digital imagery, and has high correspondence with ground-truthed field values of tree cover.
|Original language||English (US)|
|Number of pages||6|
|Journal||Photogrammetric Engineering and Remote Sensing|
|State||Published - Sep 2004|
ASJC Scopus subject areas
- Computers in Earth Sciences