Automated Multispectral Texture mapping of 3D models

Recently, thanks to the diffusion of scanning devices and the availability of powerful 3D modelling software, as well as to the improvements in the automation of the image-based modelling approach, 3D models are more and more considered in several research fields, such as inspection, navigation, object identification, visualisation and animation, and recently also in Cultural Heritage. Documentation in case of loss or damage, virtual tourism and museum, education resources, interaction (without risk of damage), and so forth are only few examples of applications where they has become a fundamental step. However, 3D models obtained with these technologies are often lacking a suitable photorealistic appearance, due to low quality acquired texture, or to the complete absence of it. Moreover, it is often of specific interest to texture with images different from photos, like multispetral/multimodal images (IR, Xrays, UV fluorescence). In such cases, a postprocessing texture mapping is needed, and it is often achieved through manual alignment of the model and the related texture. In this work, we propose a fully automatic approach for multispectral texture mapping. The method relies on the extraction from the model geometry of a depth map, in form of an image, whose pixels maintain an exact correspondence with vertices of the 3D model; the subsequent step is registratio , between such image and the chosen texture, with a very robust registration algorithm, based on Maximixation of Mutual Information. The results presented show the effectiveness of the proposed method.