Abstract
A unification framework is proposed for three-dimensional shape reconstruction using physically based models. The authors observe that most existing shape-from-X techniques use an observable (e.g., the stereo disparity, intensity, or texture gradient) and a model, which is based on specific domain knowledge (e.g., the triangulation principle, reflectance function, or texture distortion equation) to predict the observable, in 3D shape reconstruction. It is shown that all these observable-prediction-model types of techniques can be incorporated into the framework of energy constraint on a flexible, deformable image frame. Surface reconstruction is achieved through a package wrapping or a shape deformation process based on physics principles. >
Published Version
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call