The Creation of 3-D Models from a Set of Images
The generation of 3-D models is a fundamental part of computer graphics and often requires many hours of skilled labour or the use of very expensive equipment. With the increasing demand for virtual environments and objects to populate them, convenient 3-D model generation has become a very important problem to be solved. In this project we aim to provide an automated, efficient approach that does not require any specialised equipment or pre-calibration of a camera for the generation of 3-D models from a small number of views of an object.We use automatic segmentation and a two stage iterative camera calibration and volumetric intersection algorithm to produce a set of 3-D points in the form of a cloud of voxels. From this a polygonal or B-spline based surface representation of an object can be generated. The B-spline surface representation can be used to implicitly smooth the modelled surface. We extract the texture and colour of the model with reference to the original images, and obtain highly portable graphical models.
Our initial work has used a turntable to rotate the object to be modelled. However, recent work has investigated the use of a hand-held camera, so that large objects may be modelled by walking around them.
Our techniques are designed to be fast and robust, but to require no
specialised equipment and to be used by untrained users.
Staff and Students
Neill Campbell, Barry ThomasDave Gibson.