按改进的匹配算法的岩石样本图像三维重建

3D Reconstruction of Rock Specimen Images Based on Improved Matching Algorithm

为了提高岩石样本图像三维重建的准确率和重建效率,提出了基于改进的匹配算法的三维重建方法。该方法采用"回"字型分块思想,有效地提高了ORB算法中BRIEF特征描述符的学习效率,从而改善了ORB算法的匹配效果;再利用向量场一致(VFC)算法对改进的ORB算子所得的匹配对进行误匹配剔除。实验结果表明,该方法不仅能够保证岩石样本图像三维重建的精度,而且在速度上也具有较大提升。

In order to improve the accuracy and efficiency of three-dimensional （3D） reconstruction of rock images, 3D reconstruction method based on improved matching algorithm is proposed. Firstly, This method uses the ‘‘back”-shaped block thinking to improve the matching efficiency of the ORB algorithm by improving the learning efficiency of the BRIEF feature descriptor in the oriented FAST and rotated BRIEF （ ORB ） algorithm. Then the matching pairs obtained by the improved ORB operator are eliminated by using the vector field consensus （VFC） algorithm. The experimental results prove that this method can not only guarantee the accuracy of 3D reconstruction of rock specimen images, but also has a great improvement in speed.