Algorithm of Reconstruction 3D Model by Two of Its Images

The article deals with the problem of calculating cloud of points of 3D model by its images obtained by two and more cameras. Algorithm of this reconstruction is based on calculation coordinates of each points by its projection from two cameras using fact that projection of ray joining center one of these camera doesn't contain projection more one point on other camera. Such situation can be achieved by small motion of cameras. Also the problem of reconstruction 3D model by two its images is solved in the article. We consider parameterized geometric models and reconstruction problem reduce to fitting vertices of 3D model to calculated cloud of points. To solve this issue we use calculation the vectors of central moments of vertices of 3D model and cloud of points and minimization of difference between them. In addition we present algorithm of cleaning extra points which are erroneously calculated. This algorithm is based on geometric hashing of the calculated cloud of points in 3D box using images from three and more cameras.