Semantic volume texture for virtual city building model visualisation

With the rapid urbanisation and development of three-dimensional (3D) space use, space objects in residential houses are of increasing concern. Illustrating these spatial entities within a clustered and multi-layered environment is confronted with the long-standing cognitive challenges of visual occlusion, visual clutter and visual navigation. Direct illustration as cross sections and cutaways provide valuable instruction for removing occlusion while preserving global contexts to give positioning cues. However, cross-sectioning or cutting away of the popular boundary-described models suffers from computational robustness and efficiency problems, while separated boundary geometry with surface properties prevents the efficient image-based direct illustrations from implementing a visually complete, semantically consistent practice easily. This article proposes a semantic volume texture (SVT) model for direct illustration. This true-3D raster model integrates spatial pattern embedding as well, thus avoiding the costly amendments to keep semantic consistent and visually complete during illustrative cutting and reconstructing operations. The proposed model is extended to the practical base of CityGML schema, the preparation of SVT is presented and applications imitating cross sections and cutaways are demonstrated. Experiments show that SVT-based direct illustrations are effective and efficient, making the proposed model suitable for explorative visualisations in the layered micro-scale environments.