Comparison of Tree Species Classifications at the Individual Tree Level by Combining ALS Data and RGB Images Using Different Algorithms

Abstract

Individual tree delineation using remotely sensed data plays a very important role in precision forestry because it can provide detailed forest information on a large scale, which is required by forest managers. This study aimed to evaluate the utility of airborne laser scanning (ALS) data for individual tree detection and species classification in Japanese coniferous forests with a high canopy density. Tree crowns in the study area were first delineated by the individual tree detection approach using a canopy height model (CHM) derived from the ALS data. Then, the detected tree crowns were classified into four classes-Pinus densiflora, Chamaecyparis obtusa, Larix kaempferi, and broadleaved trees-using a tree crown-based classification approach with different combinations of 23 features derived from the ALS data and true-color (red-green-blue-RGB) orthoimages. To determine the best combination of features for species classification, several loops were performed using a forward iteration method. Additionally, several classification algorithms were compared in the present study. The results of this study indicate that the combination of the RGB images with laser intensity, convex hull area, convex hull point volume, shape index, crown area, and crown height features produced the highest classification accuracy of 90.8% with the use of the quadratic support vector machines (QSVM) classifier. Compared to only using the spectral characteristics of the orthophotos, the overall accuracy was improved by 14.1%, 9.4%, and 8.8% with the best combination of features when using the QSVM, neural network (NN), and random forest (RF) approaches, respectively. In terms of different classification algorithms, the findings of our study recommend the QSVM approach rather than NNs and RFs to classify the tree species in the study area. However, these classification approaches should be further tested in other forests using different data. This study demonstrates that the synergy of the ALS data and RGB images could be a promising approach to improve species classifications.

Individual Tree-Crown Detection in RGB Imagery Using Semi-Supervised Deep Learning Neural Networks

“…For example, splitting the "Tree" class into 289 living and dead trees would provide management insight when surveying for outbreaks of tree 290 pests and pathogens (Wulder et al, 2006), as well as post-fire timber operations (Vogeler et al, 291 2016). With the addition of hyperspectral data, dividing the tree class into species labels yields 292 additional insights into the economic value, ecological habitat, and carbon storage capacity for 293 large geographic areas (Deng et al, 2016). As such, deep learning-based approaches provide 294 the potential for large scale actionable information on natural systems to be derived from 295 remote sensing data.…”

“…86Deep learning has only been recently applied to airborne forestry measurements(Ayrey and 87 Hayes, 2018;Li et al, 2016). After delineating trees using LIDAR, several papers have used 88 convolutional neural networks to assign species labels to candidate trees(Deng et al, 2016; 89 Mizoguchi et al, 2017). To our knowledge, the only prior use of deep learning for vegetation 90 detection showed that passing a sliding window CNN over the entire image outperformed pixel-91 based watershed methods for detecting individual trees in palm plantations(Li et al, 2016) and 92 small shrubs in arid landscapes(Guirado et al, 2017).…”

Mapping Physiognomic Types of Indigenous Forest in New Zealand Using Space-Borne SAR, Optical Imagery and Air-Borne LiDAR

“…The PALSAR Global Mosaic [43,24] is a high-resolution slope-corrected and ortho-rectified Lband gamma-naught map. The data used for this study were acquired at 10-m resolution in a finebeam dual-polarization mode (HH and HV) for the year 2007.…”

“…In addition to imagery quality and fitness for purpose [23], the choice of classification method is also important. Nonparametric methods are proving to be effective for classifying forest types [24], especially Random Forests and Support Vector Machine (SVM) classification, due to their ability to synthesize classification functions on discrete and continuous data sets, and to deal with unbalanced data, and also to their insensitivity to noise or overtraining [25,26]. These properties enable the combination of spectral data with other types of data to improve the accuracy of classifying forest types.…”

Tree Species Classification of Broadleaved Forests in Nagano, Central Japan, Using Airborne Laser Data and Multispectral Images

“…A support vector machines (SVM) classifier that we previously identified a powerful classifier (Deng et al, 2016) was used for tree species classification. The predictive power of the models using different features was verified by five-fold crossvalidations.…”