Mangrove Growth Monitoring Based on Camera Visible Images—A Case Study on Typical Mangroves in Guangxi

Mangroves are an important coastal wetland ecosystem, and the high-throughput visible light (RGB) images of the canopy obtained by the ecological meteorological station can provide basic data for quantitative and continuous growth monitoring of mangroves. However, as for the mangroves that are subject to periodic seawater submersion, some key technical issues such as image selection, vegetation segmentation, and index applicability remain unsolved. With the typical mangroves in Beihai, Guangxi, as the object in this study, we used canopy RGB images and tidal data to find out the screening methods for high-quality nontidal submerged images, as well as the vegetation segmentation algorithms and RGB vegetation index applicability, so as to provide technical reference for the use of RGB images to monitor mangrove growth. The results showed that: 1) The critical tide levels can be determined according to the periodic changes of submersion in the mangroves, and critical tidal levels and image brightness can be used to quickly screen high-quality images of mangroves that are not submerged by seawater. 2) Machine learning and NLM filtering are effective strategies to obtain high-precision mangrove segmentation results. The machine learning algorithm has superiority in the segmentation of mangrove vegetation with a segmentation accuracy of higher than 80%, and the nonlocal mean filtering can effectively optimize the segmentation results of various algorithms. 3) The seasonal index VEG and antiseasonal index CIVE can be used as the optimal indices for mangrove growth monitoring, and the compound sine function can better simulate the change trend of various RGB vegetation indices, which is convenient for quickly judging mangrove growth changes. 4) Mangrove RGB vegetation indices are sensitive to meteorological factors and can be used to analyze the influence of meteorological conditions on mangrove growth.