TY - JOUR

T1 - Generalization of tiled models with curved surfaces using typification

AU - Guercke, Richard

AU - Zhao, Junqiao

AU - Brenner, Claus

AU - Zhu, Qing

PY - 2010

Y1 - 2010

N2 - Especially for landmark buildings or in the context of cultural heritage documentation, highly detailed digital models are being created in many places. In some of these models, surfaces are represented by tiles which are individually modeled as solid shapes. In many applications, the high complexity of these models has to be reduced for more x efficient visualization and analysis. In our paper, we introduce an approach to derive versions at different scales from such a model through the generalization method of typification that works for curved underlying surfaces. Using the example of tiles placed on a curved roof ' which occur, for example, very frequently in ancient Chinese architecture, the original set of tiles is replaced by fewer but bigger tiles while keeping a similar appearance. In the first step, the distribution of the central points of the tiles is approximated by a spline surface. This is necessary because curved roof surfaces cannot be approximated by planes at large scales. After that, the new set of tiles with less rows and/or columns is distributed along a spline surface generated from a morphing of the original surface towards a plane. The degree of morphing is dependent on the desired target scale. If the surface can be represented as a plane at the given resolution, the tiles may be converted to a bump map or a simple texture for visualization. In the final part, a perception-based method using CSF (contrast sensitivity function) is introduced to determine an appropriate LoD (level of detail) version of the model for a given viewing scenario (point of view and camera properties) at runtime.

AB - Especially for landmark buildings or in the context of cultural heritage documentation, highly detailed digital models are being created in many places. In some of these models, surfaces are represented by tiles which are individually modeled as solid shapes. In many applications, the high complexity of these models has to be reduced for more x efficient visualization and analysis. In our paper, we introduce an approach to derive versions at different scales from such a model through the generalization method of typification that works for curved underlying surfaces. Using the example of tiles placed on a curved roof ' which occur, for example, very frequently in ancient Chinese architecture, the original set of tiles is replaced by fewer but bigger tiles while keeping a similar appearance. In the first step, the distribution of the central points of the tiles is approximated by a spline surface. This is necessary because curved roof surfaces cannot be approximated by planes at large scales. After that, the new set of tiles with less rows and/or columns is distributed along a spline surface generated from a morphing of the original surface towards a plane. The degree of morphing is dependent on the desired target scale. If the surface can be represented as a plane at the given resolution, the tiles may be converted to a bump map or a simple texture for visualization. In the final part, a perception-based method using CSF (contrast sensitivity function) is introduced to determine an appropriate LoD (level of detail) version of the model for a given viewing scenario (point of view and camera properties) at runtime.

KW - Building Models

KW - Curved

KW - Generalization

KW - Multi-Scale Representation of Spatial Data

KW - Roof

KW - Tiles

UR - http://www.scopus.com/inward/record.url?scp=84923645299&partnerID=8YFLogxK

M3 - Conference article

AN - SCOPUS:84923645299

VL - 38

SP - 39

EP - 44

JO - International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences - ISPRS Archives

JF - International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences - ISPRS Archives

SN - 1682-1750

T2 - Joint International Conference on Theory, Data Handling and Modelling in GeoSpatial Information Science

Y2 - 26 May 2010 through 28 May 2010

ER -