Analysis of the accuracy of pixel equivalent method and SfM method in estimating diameter at breast height, tree height, and volume of street trees
摘要
To compare the measurement accuracy of Structure from Motion (SfM) photogrammetry and the pixel equivalent method in acquiring urban street tree parameters, the diameter at breast height (DBH), tree height, and volume of 56 trees from three species (Taxodium distichum, Ginkgo biloba, and Quercus rubra) were estimated using smartphone image data. The results showed that both methods performed well for DBH estimation, with the pixel equivalent method showing slightly higher goodness of fit (Bias = 0.1241 cm, RMSE = 0.6511 cm, R2 = 0.984), whereas smartphone-based SfM showed slightly lower overall error(Bias = -0.6664 cm, RMSE = 1.3543 cm, R2 = 0.938). However, smartphone-based SfM exhibited substantially lower accuracy for tree height (Bias = 0.6973 m, RMSE = 2.5124 m, R2 = 0.017) and volume estimation (Bias = 0.0384 m3, RMSE = 0.0582 m3, R2 = 0.896). In contrast, the pixel equivalent method demonstrated superior performance for tree height (Bias = 0.2935 m, RMSE = 0.6653 m, R2 = 0.713) and volume (Bias = 0.0168 m3, RMSE = 0.0423 m3, R2 = 0.931). Overall, the pixel equivalent method produced smaller systematic bias, lower RMSE, and higher model fit, outperforming smartphone-based SfM under the present conditions, suggesting that it is more suitable for estimating urban street tree parameters in this study.