Tree Stem Diameter Estimation From Volumetric TLS Image Data

Abstract

Abstract:Recently, a new method on tree stem isolation using volumetric image data from terrestrial laser scans (TLS) has been introduced by the same authors. The method transfers TLS data into a voxel grid data structure and isolates the tree stems from the overall forest vegetation. While the stem detection method yields on a three dimensional localisation of the tree stems, the present study introduces a supplemental technique, which accurately estimates the diameter at breast height (DBH) from the stem objects. Often, large pieces of the stems are occluded by other vegetation and are only partially represented in the laser scanning data, not covering the complete circumference. Therefore, it was not possible to measure the diameter at 130 cm height directly on the stem imagery. Instead, a method has been developed, which estimated the diameter from the fragmented stem information at the specific cross sections. The stem information was processed in a way, which allowed applying a Hough transform to the image for fitting circles to the cross sections. In contrast to other studies, Hough transform was applied to single stem images with information from other vegetation parts already being removed. Even in cases where only a single and very small fragment of a stem is available, the diameter could be estimated from the curvature. It also has been demonstrated that the image resolution for DBH measurement can be significantly higher than the resolution used for stem isolation in order to increase the precision. Verification of the computed DBH on nine spatially independent test sites showed that applying the Hough transform to single stem cross section images produced accurate results. When excluding the five strongest individual outliers a bias of −0.02 cm, a root mean square error (RMSE) of 2.9 cm and a R 2 of 0.98 were achieved.

Impact of Beam Diameter and Scanning Approach on Point Cloud Quality of Terrestrial Laser Scanning in Forests

“…Another problem occurring with large laser beam diameters is that the object shape tends to be flattened. This could lead to a bias in diameter estimation, e.g., when applying the widely used Hough transform approach (see [33]- [36]) to fit circles or cylinders in the point cloud.…”

Comparing Three Different Ground Based Laser Scanning Methods for Tree Stem Detection

“…One reason for the blurry shapes describing the smaller trees might be the point cloud precision. Another reason might be a change in the stem inclination due to wind conditions [21]. All laser scanners had problems to detect the smaller trees.…”

“…A root mean square error (RMSE) of 2.4 cm was obtained when estimating the DBH. Based on a novel method for tree stem identification, Heinzel and Huber [21] used TLS to estimate DBH, and obtained an RMSE of 2.9 cm. In a study by Bauwens et al [15], three systems based on TLS and MLS were compared on trees with DBH > 10 cm.…”

Identifying Tree-Related Microhabitats in TLS Point Clouds Using Machine Learning

“…Over the past decade, the potential of TLS has been actively tested to complement or, in the future, even replace extensive field surveys for forest monitoring, including for conventional NFIs [14][15][16][17]. Researchers have already made big advances in defining optimal scanning designs for inventory plots [18] and for deriving a wide range of plot-level and tree-level characteristics, such as tree-stem detection [19] and modeling [20][21][22], tree species recognition [23,24], DBH extraction [25], determination of leaf area distribution [26], volume estimation [27], detection of clear wood content [28] and wood defects [29], assessment of timber quality and timber assortments [30]. TLS 3D point clouds have also successfully been used to derive forest structural information to evaluate the suitability of a forest stand as a habitat for animals, e.g., bats [31] or birds [32].…”