基于SIFT特征检测和三次插值样条的快速曲面重构算法

Fast Mesh Reconstruction Algorithm Based on SIFT Features Detection and Cubic Spline Curve Fitting

摘要:

针对传统的曲面重构算法一直存在重构精度低、效率差的问题,提出新的快速曲面重构算法；基于三次插值样条和改进SIFT特征检测和匹配方法,先对二维图像预处理后,检测其边界点并且拟合,然后将区域按四个象限进行分区,用三次插值样条边界进行曲面重构,同时结合改进的SIFT特征检测和匹配方法,检测各个区域曲面边界的特征点,然后再进行匹配拼接；实验结果表明,该算法得到曲面模型的时间效率有很大的提高,并且曲面效果好,光滑、平顺。

Abstract:

Aiming at the problem of low reconstruction precision and poor efficiency, a new method of fast surface reconstruction is proposed. Based on the three spline interpolation and improved SIFT feature detection and matching method, first to two-dimensional image after pretreatment, detection of the boundary points and fitting, and then the area is partitioned by four quadrants, surface reconstruction with three interpolation spline boundary feature points, combined with improved SIFT feature detection and matching method detection of each region boundary surface, and then matching stitching. Experimental results show that the proposed algorithm can improve the time efficiency of the surface model, and the surface effect is good, smooth and smooth.