A multiple representation data structure of 3D building textures

Texture is an important element for visualization of 3D city models and often takes a big proportion of the total data volume. In order to simplify 3D city model textures for dynamic visualization in different scales, a multiple representation data structure, TextureTree is proposed to store building textures in different LoDs. First, the texture image is iteratively segmented by horizontal or vertical dividing zone (edge or background from edge detection) until each sections are basically in the same color. Then textures in all sections are represented by their main color and the TextureTree is created based on the color difference between the adjacent sections. With the TextureTree, the simplified texture in different LoDs can be dynamically generated. The experiment results show that the data volume of building textures can be reduced by TextureTree while the required visual similarity is preserved.