Shoreline_veget-A new shoreline extraction method for vegetation seas, with high robustness, accuracy and scalability

Weihua Li ^* lianglin liu wei zhu jiuming li teng liu

• Jinggangshan University, Ji'an, China

To address the issue of vegetation obstructing water bodies and resulting in missing information in vegetation sea areas, existing methods that focus on various types of shorelines often exhibit limited algorithm stability and accuracy. This study introduces a method, termed Shoreline_veget, for solving the problem of missing remote sensing information due to vegetation occlusion, improving the accuracy of shoreline extraction, and finally giving a shoreline extraction model for all coasts. The method comprises four main modules: The first module involves regenerating the point cloud for the vegetation sea area and classifying the vegetation and land. The second module extracts the precise boundary between vegetation and land, projects this vegetation boundary onto the water surface, corrects the elevation, applies offsets, and then merges the two boundaries. The third module integrates the merged boundary into the coastal zone point cloud, then designs an elevation gradient function based on the terrain and shoreline direction to determine the attributes of each point within the merged boundary, and accurately adjusts each point based on these attributes. The fourth module finalizes the merged boundary and utilizes tidal data and shoreline correction models to compute the shoreline. The experiments demonstrate that this method significantly enhances the accuracy of shoreline. Specifically, it reduces the overall shoreline accuracy in data one from 0.6658, 0.3854, and 0.4127 (as observed with three comparative methods) to 0.1531. Compared to the least accurate method, this method improved the overall shoreline accuracy by 0.5127 m. The result confirm that the proposed method offers superior stability, and this methodology provides new technology to measure, map, and manage shorelines, offers valuable insight for related research.