Monocular three-dimensional self-calibrating surface reconstruction using digital photogrammetry

Photogrammetry, which has significantly matured throughout the years, is an optical technique to accurately measure the depth of spatial points from their images. In this paper, a self-calibrating measurement technique based on photogrammery is presented. A digital camera with zoom lenses is used to capture images of the investigated object. The image sequence is then transferred to a Matlab® code to detect the interest points, then to compute the 3-D coordinates of those interest points and the camera poses by matrix factorization operation. Neither calibration object nor a priori information of the camera is needed. The proposed 3-D reconstruction technique is proved able to measure the 3-D shapes of various surfaces with the error of 1 part in 1000 of their sizes.