Automatic acquisition of planar-faced virtual models

An algorithm for fast automatic acquisition of virtual models of planar-faced textured objects from registered range and colour images is introduced. The algorithm consists of four main steps-range image segmentation, texture analysis, virtual model geometry inference and texture synthesis. An efficient and robust range image segmentation is based on the combination of range profile modelling and line-based region growing. The texture is modelled using a multi-resolution Markov random field based model.