Euclidean Reconstruction By Means Of An Uncalibrated Structured Light Sensor

Abstract

A method of uncalibrated reconstruction through structured lighting is presented in this paper. The co-ordinates of 3-D points and the projection matrices are simultaneously determined in a parameter estimation approach. The reconstruction is performed in a projective frame, up to a projective transformation, by only using pixel correspondences. It is shown that an Euclidean reconstruction can be recovered by constraining the projective transformation by using geometrical knowledge about the scene. Moreover, it is described how the pattern projection is used to grab this kind of knowledge. This method provides our sensor with adaptive capabilities and permits to be used in mobile robotics. Experimental results realised both on simulated and real data are presented.