3D reconstruction using a spherical spiral scan camera
Construction of 3D models representing industrial products/objects is commonly used as a preliminary step of production process. These models are represented by a set of points which can be combined by planar patches (i.e. triangulation) or by smooth surface approximtions. In some cases, we may need to construct the 3D models of real objects. This problem is known as the 3D reconstruction problem which is one of the most important problems in the field of computer vision. In this thesis, a sytem has been developed to transform images of real objects into their 3D models automatically. The system consists of a PC, an inexpensive camera and an electromechanical component. The camera attached to this component moves around the object over a spiral trajectory and observes it from different view angles. At the same time, feature points of object surface are tracked using a tracking algorithm over object images. Then tracked points are reconstructed in 3D space using a stereo vision technique. A surface approximation is fitted to this 3D point set as a last step in the process. Open source .Intel®-OpenCV. C library is used both in image capturing and in image processing.