Diameter at breast height is a standard method for characterizing trees and ultimately forests over large geographical scales. Expressing the cross-section of a tree stem with a diameter measurement has inherent errors, as stems are not perfect conical frustums. Diameter suffices as an approximation, since methods for the exact cross-sectional area of a stem are infeasible in practice. There are other primary sources of error in diameter at breast height measurements: diameter measurement error and breast height measurement error. This paper addresses the latter in the context of an automated measurement system using 3D photography. We consider breast height estimation a fundamental component in such systems, as errors in measurement height can precipitate significant errors in stand-level attribute estimation as diameter changes along the height of the stem because of taper. We evaluate the efficacy of breast height estimation using stereo photogrammetry by fitting a planar surface to the ground and elevating the plane to breast height. Results from a validation test on 560 observations show that breast height can be estimated with a root mean squared error of 10.20 cm within 10 m from the camera and 13.36 cm within 20 m from the camera on flat terrain in varying levels of stem density.

中文翻译：

---

估计立体图像中的乳房高度的地平面检测评估

胸高直径是表征大地理范围上的树木以及最终森林的标准方法。用直径测量值来表示树茎的横截面存在固有的误差，因为茎不是理想的圆锥形截头圆锥体。直径大约是足够的，因为在实践中，无法精确计算出杆的确切横截面积。乳房高度测量中的直径误差还有其他主要来源：直径测量误差和乳房高度测量误差。本文在使用3D摄影的自动测量系统中解决了后者。我们认为乳房高度估算是此类系统的基本组成部分，因为随着锥度的增加，直径沿茎杆的高度变化，因此测量高度的误差可能会在站立水平属性估计中引发重大误差。我们通过将平面贴合到地面并将平面提升至乳房高度，使用立体摄影测量法评估乳房高度估计的功效。对560个观测值进行的验证测试的结果表明，在平坦的地形上，在不同茎密度水平下，可以估计距摄像头10 m以内10.20 cm和距摄像头20 m以内13.36 cm的均方根误差。