Three-dimensional reconstruction of binocular stereo vision based on im- proved SURF algorithm and KD-Tree matching algorithm

With the constant development of computer science and technology, binocular stereo vision as a special form of computer vision plays a vital role in implementing computer change detection, image correction, three-dimensional reconstruction, and is widely applied in computer vision fields such as aerial mapping, visual navigation, motion analysis and industrial inspection. However, as it is difficult to perform binocular stereo vision, and precise parallax principle and mathematical method are required. Thus this study rapidly and effectively realizes three-dimensional reconstruction of binocular stereo vision by feature description vector generated by improved surface (SURF) algorithm combined with K-dimension tree (KD-Tree) searching. Effective combination of improved SURF algorithm and KD-Tree significantly enhances sense of reality of three-dimensional scene. Subject Categories and Descriptors I.3.7 [Three-Dimensional Graphics and Realism]: Threedimensional Reconstruction; I.6 [Simulation and Modeling]: Vision Model General Terms: Gamma correction, Three-dimensional reconstruction