Abstract: 3D modeling based on oblique photogrammetry has become the development trend of 3D urban construction. The vegetations with uneven surface need a large number of triangles to depict in the urban model, but this is not the focus of 3D real scene data recording and expression, it is difficult to display and apply the model due to a large amount of data. So, this paper proposes a 3D model simplification method of oblique photography considering the types of ground objects. Firstly, vegetation area is extracted by using Markov random field (MRF) based on textural and geometric information. Then, the quadric error metrics (QEM) algorithm is used to simplify the vegetation area. Finally, the texture remap of the simplified white film model is performed. The experimental result shows that the method in this paper can extract and simplify the vegetation area effectively, the texture remap result is almost the same as the original model in appearance. The method proposed in this paper has achieved the desired result.

Key words: oblique photogrammetry, 3D real scene model, vegetation extraction, Markov random field, mesh simplification