Three Dimensional Model Reconstruction of a Cuboid Based on Stereo Vision

Abstract

Three-dimensional reconstruction of objects is one of the key elements in production processes. Stereo vision is one of the robost techniques in this field. Accuracy of the reconstruction is influenced by several parameters that mainly determined through the calibration process. Besides, symetring of two accquired images effects on the accuracy. In the present research, a three-dimensional model reconstruction system based on stereo vision technique included a camera, a robot, a lightening system, a cubic object, and IMAQ Image Processing and LabVIEW Softwares. The inactive stereo vision was considered such that the images were acquired by the same camera in two different positions. The cuboid model was implemented after camera calibration. The corners of the object were considered to remove the symmetry problems. After comparing the dimensions of the reconstructed model with those of real objects, the maximum error value of the system was obtained as 0.75 mm and the error variance as 0.09 mm.