In this paper

the methods for the uncalibrated two-views 3D reconstruction is proposed. The methods employ geometric constraints available from geometric relationships that are plentiful in manmade structure-such as parallelism and orthogonality of lines and planes

these constraints lead to simple method to calibrate the intrinsic parameters of the camera. This is done by determining the vanishing points associated with parallel lines in the world

under the assumption of zero skew and known aspect ratio

three mutually orthogonal directions are exploited to give the camera calibration matrix . It is possible to obtain only a projective reconstruction from the fundamental matrix. If each image is calibrated

it be able to convert from the uncalibrated fundamental matrix to the essential matrix. A Euclidean reconstruction would be preferable. Once the essential matrix is recovered

if the first camera is assumed to be at origin of the coordinate system

then it is a simple matter to calculate the rotation and translation of the second camera relative to the first. After the camera intrinsic and extrinsic parameters had been estimated

camera projection matrices may be recovered and used to estimate the structure. The 3D reconstruction process has two stages: the first to recover the camera positions and motions

the second step involves triangulation to recover the 3D points. The validity of the proposed algorithm is confirmed by experiment for a number of multi-plannar scenes. The reconstructed scene is modeled. New images are generated of the model for new view points. The geometry agrees with our perception of scene. The angle between the two reconstructed planes looks just like a right angle.