Extracting the Forest Type From Remote Sensing Images by Random Forest

• Published in: IEEE sensors journal Vol. 21; no. 16; pp. 17447 - 17454
• Main Authors: Linhui, Li, Weipeng, Jing, Huihui, Wang
• Format: Journal Article
• Language: English
• Published: New York IEEE 15.08.2021; The Institute of Electrical and Electronics Engineers, Inc. (IEEE)
• Subjects: Accuracy; Classification; Color imagery; Decision trees; Feature extraction; Forest type extraction; Forestry; Identification methods; Image resolution; Image segmentation; object oriented; Random forests; Remote sensing; RF classification; Sensors; Spatial data; Support vector machines; SVM classification; Vegetation; Vegetation index; Vegetation mapping
• ISSN: 1530-437X; 1558-1748
• DOI: 10.1109/JSEN.2020.3045501

Identifying the types of forest and the corresponding distribution is of significance in forest resource monitoring and management. Considering the low accuracy of extracting the information of forest types from high-resolution remote sensing images and the lack of an effective identification method. The GF-2 remote sensing image in the Laoshan construction area of the Maoershan Forest Farm, Heilongjiang Province was as the data source, supplemented by aerial RGB images with a resolution of 0.2 m and the second type inventory of forest resources data. Considering the spatial characteristics of the spectrum, texture, vegetation index, terrain, multiscale segmentation was performed, the optimal feature space was constructed, and the number of decision trees was estimated. In this manner, an object-oriented random forest (RF) scheme was established. Comparative experiments were performed using the support vector machine(SVM) classifier. The experimental results indicated that the overall accuracy and kappa coefficient of the proposed method was 83.16% and 79.86%, respectively, higher than those of the SVM classification method. These findings demonstrated that the proposed method can effectively increase the classification accuracy of forest types.