Error Analysis of Euclidean Reconstruction from Translational Motion

Abstract

Euclidean Reconstruction of an object is one of the challenging problems in computer vision. This paper uses the singular value decomposition (SVD) method to solve the problem of Euclidean Reconstruction from translational motion, under one calibrated camera. This paper mainly analyzes how the different elements (camera intrinsic parameters, corresponding error, the value and the direction of translation) affect the reconstruction results. The error analysis is based on both the simulated data experiments and the real data experiment. The results of experiments show that the corresponding error, the value and the direction of translation are the main elements affect the reconstruction results.