Automatic Generation Method of Vegetation Elements in Urban Traffic 3D Models Based on Street View Recognition

An automatic generation method for vegetation elements in urban traffic 3D models based on street view recognition is proposed. It aims to address issues in vegetation 3D modeling, such as missing elements, high labor costs, difficulties in material integration, and insufficient automation capabilities, through an integrated process including standardized material collection, vectorized matching, and procedural generation. First, vegetation categories and their typical attributes are defined, 3D models and street view images are collected in a targeted manner, and metadata is unified and stored in a database. Further, based on semantic segmentation and fine-grained recognition networks, vegetation recognition in street view images is realized. At the same time, candidate sets are screened based on ontology mapping, and the cosine similarity vector retrieval method is used to select the most matching vegetation model by comparing the local feature vectors of the recognized ROIs (Regions of Interest) with the pre-stored feature vectors in the material library.