Automatic Forest DBH Measurement Based on Structure from Motion Photogrammetry

Measuring diameter at breast height (DBH) is an essential but laborious task in the traditional forest inventory; it motivates people to develop alternative methods based on remote sensing technologies. In recent years, structure from motion (SfM) photogrammetry has drawn researchers’ attention surveying for its economy and high precision as light detection ranging (LiDAR) are always expensive. This study explores automatic DBH measurement method SfM. Firstly, we proposed a new image acquisition technique that could reduce number of images accuracy measurement. Secondly, developed estimation pipeline sample consensus (RANSAC) cylinder fitting with Least Median Squares impressive speed comparable LiDAR. For application SfM survey, graphical interface software Auto-DBH integrated reconstruction was developed. We sampled four plots different species verify performance method. The result showed first two plots, where trees’ stems were good roundness, root mean squared error (RMSE) 1.41 cm 1.118 relative 4.78% 5.70%, respectively. third plot’s damaged trunks low roundness reduced RMSE 3.16 10.74%. average rate trees 91%. Our procedure relatively fast takes only 2 s estimate tree, which much more rapid than direct physical measurements tree trunk diameters. proves reach accuracy, close terrestrial laser scanning (TLS) plot scale successful indicates promising inventory.