基于SIFT特征匹配算子的三维重建方法研究

Research on the 3D Reconstruction Method Based on Scale-invariant Feature Transform Feature Matching Algorithm

为了改善基于双目立体视觉的逆向工程中零件三维重建的精度,提出了一种基于尺度不变特征变换(SIFT)算子的三维重建方法。该方法通过SIFT算子搜索得到两幅对应的数字图像中的匹配特征点,并利用标定好的摄像机参数将图像特征点映射成空间直线,最后采用异面直线公垂线中点逼近法求取三维空间点的坐标。该方法不需要额外的设备来辅助特征点的匹配,实验结果表明该方法具有良好的鲁棒性。

In order to improve the precision of the 3D reconstruction for parts in reverse engineering based on binocular stereoscopic vision,the method of 3D reconstruction based on scale-invariant feature transform（SIFT） algorithm is proposed.With this method,through SIFT algorithm,two of the matching pixel points in corresponding digital images are found;and the image feature points are mapped into space lines using calibrated camera parameters.The coordinates of the three dimensional points are derived with approximation of the middle point of the common perpendicular line in different planes.This method does not require extra equipment to assist feature matching.The result of experiments shows that the method offers excellent robustness.