Extraction Methods of Forest Canopy Closure Based on Tencent Street View

The study extracted the panoramic image set from Tencent Street View, transformed the cylindrical panorama into azimuth fish-eye image through image mosaic technology, and combined image enhancement, image binarization and other digital image processing techniques to extract the canopy closure of forest in Shanghai Botanical Garden. The extraction results were compared with the canopy closure calculated by the unsupervised classification method and the canopy closure of the field surveys to verify the feasibility and accuracy of the method. The results show that the absolute error mean (0.027) and standard deviation (0.031) of Tencent Street View method are smaller than the unsupervised classification method, which indicates that the absolute error dispersion of Tencent Street View method is small and relatively stable. The R² of Tencent Street View is 0.977, and the RMSE is 0.054, which indicates that Tencent Street View is a better estimation method. The slope of the fitted line of Tencent Street View is 0.952, which indicates that the overall estimated value of Tencent Street View is smaller; the slope of the fitted line of unsupervised classification method is 1.013, indicating that the overall estimated value of unsupervised classification method is larger.