Integration of Two Cameras and Kinect Range Sensor for Three-Dimensional Data Reconstruction

摘要

In applying a stereo vision system for 3D data reconstruction, stereo matching are required to triangulate coordinates of data cloud. In this work, we developed a two-dimensional matching algorithm based on the separable successive weighted summation (SWS) method to simplify the intensity-dependent aggregation procedure in the matching. Besides, we integrate the depth information provided by the Microsoft Kinect sensor with that provided by a camera pair to improve the percentage of correct corresponding locations. Experimental results with several real scenes show that the proposed method can significantly improve the quality and accuracy of stereo matching.