基于改进SIFT立体匹配算法的双目三维重建研究

Research on Binocular 3D Reconstruction Based on Improved SIFT Stereo Matching Algorithm

针对双目三维重建视觉匹配中图像视差计算和图像像素匹配效率低这一问题,提出一种改进SIFT立体匹配算法。采用自适应模糊替换经典SIFT算法中的高斯模糊以解决高斯函数计算资源需求过大、时间过长的问题。对比不同算法处理的图片,经改进的算法匹配率在91%左右,符合要求并随着图片分辨率的提高运行时间减少。

Aiming at the low efficiency of image disparity calculation and image pixel matching in binocular 3D reconstruction visual matching,an improved SIFT stereo matching algorithm is proposed.The adaptive blur is used to replace the Gaussian blur in the classic SIFT algorithm to solve the problem of excessive demand for Gaussian function calculation resources and long time.Comparing pictures processed by different algorithms,the improved algorithm has a matching rate of about 91%,which meets the requirements and reduces the running time as the picture resolution increases.