Automated Multispectral Texture mapping of 3D models
Authors: Pelagotti A., Del Mastio A., Uccheddu F., Remondino F.
Autors Affiliation: CNR – Istituto Nazionale di Ottica Applicata; Department of Electronics and Telecommunications, University of Florence, Italy; 3D Optical Metrology Group, Bruno Kessler Foundation (FBK), Trento, Italy
Abstract: Recently, thanks to the diffusion of scanning devices and theavailability of powerful 3D modelling software, as well as tothe improvements in the automation of the image-basedmodelling approach, 3D models are more and more consideredin several research fields, such as inspection, navigation,object identification, visualisation and animation, andrecently also in Cultural Heritage. Documentation in case ofloss or damage, virtual tourism and museum, education resources,interaction (without risk of damage), and so forthare only few examples of applications where they has becomea fundamental step.However, 3D models obtained with these technologies areoften lacking a suitable photorealistic appearance, due tolow quality acquired texture, or to the complete absence ofit.Moreover, it is often of specific interest to texture with imagesdifferent from photos, like multispetral/multimodal images(IR, Xrays, UV fluorescence). In such cases, a postprocessingtexture mapping is needed, and it is oftenachieved through manual alignment of the model and therelated texture. In this work, we propose a fully automaticapproach for multispectral texture mapping. The methodrelies on the extraction from the model geometry of a depthmap, in form of an image, whose pixels maintain an exactcorrespondence with vertices of the 3D model; the subsequentstep is registratio , between such image and the chosentexture, with a very robust registration algorithm, basedon Maximixation of Mutual Information.The results presented show the effectiveness of the proposedmethod.
Conference title: European Signal Processing Conference 2009
Place: Glasgow, UK
KeyWords: image processing; 3D modeling; automated texturing