Recently, different approaches for uncalibrated stereo have been suggested which permit projective reconstructions from multiple views. These use weak calibration which is represented by the epipolar geometry, and so we require no knowledge of the intrinsic or extrinsic camera parameters. In this paper we consider projective reconstructions from pairs of views, and compare a number of the available methods. Projective stereo algorithms can be categorized by the way in which the 3D coordinates are computed. The first class is similar to traditional stereo algorithms in that the 3D world geometry is made explicit; the initial phase of the processing always involves the estimation of the camera matrices from which the 3D coordinates are computed. We show how the camera matrices can be computed either from point correspondences, or how they are constrained by the fundamental matrices. The second class of algorithms are based on implicit image measurements which are used to compute project...

We present a novel technique for calibrating a binocular stereo rig by using the information from both scenes and classical calibration objects. The calibration provided by the calssical methods is only valid for the space near the position of the calibration object. Our technique takes the advantage of the rigidity of the geometry between two cameras. The idea is to first estimate precisely the epipolar geometry which is valid for a wide range in space from all available matches. This allows to conduct a projective reconstruction. Using the a priori knowledge of the calibration object, we are eventually able to calibrate the stereo rig in a Euclidean space. The proposed technique has been tested with a number of real images, and significant improvement has been observed.