Rapid Tree Diameter Computation with Terrestrial Stereoscopic Photogrammetry

Abstract

Forest inventorying is time-consuming and expensive. Recent research involving photogrammetry promises to reduce the cost of inventorying. Existing photogrammetry methods require substantial data-processing time, however. Our aim was to reduce data-acquisition and processing times while obtaining relatively accurate diameter estimates compared to manual and other digital measurements. We developed an algorithm to identify the ground and measure diameter at breast height (dbh) or any height along a stem from the recorded video footage of trees taken with a stereo camera. Footage acquisition time, dbh root mean square error, and mean absolute error were used as comparison metrics with other methods. The time to perform three recordings for 40 trees was about 30 minutes. We recorded data at 1 m, 3 m, and 5 m from the trunk, and our dbh root mean square errors were 1.28 cm (0.50 in.), 1.47 cm (0.58 in.), and 2.57 cm (1.01 in.), respectively, using manual measures as the control. This terrestrial stereoscopic photogrammetric method is much more efficient computationally than popular terrestrial structure-from-motion photogrammetry and substantially lowers time, costs, and complexity for data acquisition and processing compared with terrestrial laser scanning.

forest inventory, stereoscopic photogrammetry, tree diameter, depth image, TSP