Dynamic object recognition method for tree obstacle hidden dangers based on LOD point cloud rendering and image segmentation technology

There are small holes and inconsistencies in the foreground of the tree obstacle hidden danger image, resulting in complex and diverse point cloud or image data, making it difficult to fully confirm the distance between vegetation and towers, and errors in the location of tree obstacle hidden danger markers. Therefore, a dynamic target recognition method for tree obstacle hidden dangers based on LOD point cloud rendering and image segmentation technology is proposed. Apply LOD detail level point cloud rendering to render laser point cloud data. Using deep learning algorithms to segment tree obstacle hidden image features. Build a dynamic target recognition model for identifying hidden dangers in transmission line channel trees, and identify hidden dangers in transmission line channel trees. The experimental test results show that the method proposed in this article can comprehensively identify tree obstacle hidden dangers, and the identification mark positions are very accurate.