Three-dimensional Reconstruction of Objects Obtained by Two Orthogonal Cameras

This paper presents a technique for 3D reconstruction of objects from a sequence of images acquired with two cameras. The cameras are mounted so that their axes remain orthogonal, both toward the subject. A mechanical device, controlled in angular position, allows movement of the cameras as they acquire an image sequence. From the images acquired different views of the silhouette of the object are extracted. The silhouettes processing to generate a 3D point cloud and generate a mesh using Delaunay triangulation techniques. Controlling the angular position of the cameras, the algorithm for the image processing and the three-dimensional reconstruction is performed using a GUI developed in Visual C++. The results of the reconstruction process are modeled in a virtual environment to get a real view of the object.