3D Reconstruction: Novel Method for Finding of Corresponding Points using Pseudo Colors

Abstract

This paper deals with the reconstruction of spatial coordinates of an arbitrary point in a scene using two images scanned by a 3D camera or two displaced cameras. Calculations are based on the perspective geom-etry. Accurate determination of corresponding points is a fundamental step in this process. The usually used methods can have a problem with points, which lie in areas without sufficient contrast. This paper describes our proposed method based on the use of the relationship between the selected points and area feature points. The proposed method finds correspondence using a set of feature points found by SURF. An algorithm is proposed and described for quick removal of false correspondences, which could ruin the correct reconstruction. The new method, which makes use of pseudo color image representation (pseudo coloring) has been proposed subsequently. By means of this method it is possible to significantly increase the color contrast of the surveyed image, and therefore add more information to find the correct correspondence. Reliability of the found correspondence can be verified by reconstruction of 3D position of selected points. Executed experiments confirm our assumptions