Leaf Segmentation Based on k-Means Algorithm to Obtain Leaf Angle Distribution Using Terrestrial LiDAR

Abstract

It is critical to take the variability of leaf angle distribution into account in a remote sensing analysis of a canopy system. Due to the physical limitations of field measurements, it is difficult to obtain leaf angles quickly and accurately, especially with a complicated canopy structure. An application of terrestrial LiDAR (Light Detection and Ranging) is a common solution for the purposes of leaf angle estimation, and it allows for the measurement and reconstruction of 3D canopy models with an arbitrary volume of leaves. However, in most cases, the leaf angle is estimated incorrectly due to inaccurate leaf segmentation. Therefore, the objective of this study was an emphasis on the development of efficient segmentation algorithms for accurate leaf angle estimation. Our study demonstrates a leaf segmentation approach based on a k-means algorithm coupled with an octree structure and the subsequent application of plane-fitting to estimate the leaf angle. Furthermore, the accuracy of the segmentation and leaf angle estimation was verified. The results showed average segmentation accuracies of 95% and 90% and absolute angular errors of 3° and 6° in the leaves sampled from mochi and Japanese camellia trees, respectively. It is our conclusion that our method of leaf angle estimation has high potential and is expected to make a significant contribution to future plant and forest research.

Apple Shape Detection Based on Geometric and Radiometric Features Using a LiDAR Laser Scanner

“…Brodu and Lague [29] proposed that the set of points within the spherical neighborhood can be utilized to define whether the geometric structure of points is closer to a line, plane surface, or a random cluster. Several researchers in forestry applications used the method to include reflectance intensity (R ToF ) and geometric features such as the linearity to discriminate foliage from woody parts [30,31] or to estimate the leaf angle [32]. Whereas in agriculture, Gene-Mola et al [33] proposed a methodology that combines the backscattered reflectance intensity and a geometric factor for spherical shapes based on the eigenvalues of clusters in apple trees to localize the fruits.…”

Development and Validation of a LiDAR Scanner for 3D Evaluation of Soil Vegetal Coverage

“…In other recent papers, LiDAR technology was used in combination with digital cameras [31], multispectral imaging (MI) [32,33] or hyperspectral imaging (HI) [33,34] for different purposes that require high resolution models for measuring very large areas or volumes. In some cases, terrestrial lidar system (TLS) technology was used [33,35,36] and in other cases airborne sensors were mounted on Unmanned Aerial Vehicles (UAV) [32,34].…”

3D Point Clouds in Forest Remote Sensing

“…Table 1 provides an overview of the research articles included in the SM, summarizing the country of the study area, the RS data type, the scale, the variable of interest and the main methods used. [16] and one technical note [3]. Among these, only one paper exclusively uses passive RS data [21], while 29 papers use at least one LiDAR dataset in the analysis [1][2][3][4][5][6][7][8][9][10][11][12][13][14][15][16][17][18][19][20][22][23][24][25][26][27][28][29][30].…”

“…Among these, only one paper exclusively uses passive RS data [21], while 29 papers use at least one LiDAR dataset in the analysis [1][2][3][4][5][6][7][8][9][10][11][12][13][14][15][16][17][18][19][20][22][23][24][25][26][27][28][29][30]. Ten papers exclusively use airborne laser scanning (ALS) data [4,6,7,10,11,13,18,23,26,27], nine papers exclusively use terrestrial laser scanning (TLS) data in the analysis [3,9,15,16,20,22,24,25,30], two papers exclusively use mobile laser scanning (MLS) data …”

“…Finally, five papers use combined active and passive remote sensing data sets [2,14,17,19,28]. Regarding the scale of the analysis, 18 of the studies perform individual tree level (ITL) analysis [1][2][3][8][9][10][11][12][14][15][16]19,20,[23][24][25][26]30], eight papers report stand level (SL) analysis [6,7,17,18,21,22,27,29] and four report a combination of ITL and SL [4,5,13,28]. Tree position, diameter at breast height (DBH) and individual tree height (h) are the most common variables of interest, analyzed in nine, six and six papers, respectively, while the most commonly used methods are 3D reconstruction, point filtering and statistical modelling, which are used in eight, five and five papers, respectively (see Table 1).…”