三焦点张量的最小代数误差非迭代算法研究

A Non-iterative Algorithm for Trifocal Tensor of Algebraic Minimization Error

摘要:

通过普通数码相机获得的3幅未标定图像,采用RANSAC算法进行两两匹配获得3幅图像的一组兴趣点对,利用最小代数误差非迭代算法在MATLAB中计算出三焦点张量,提出一种最小代数误差非迭代算法。实验结果表明,该方法是一种有效的三焦点张量计算方法,计算复杂度小、实现效率高。能够得到较准确的三焦点张量,为下一步相机自动标定以及三维场景的自动重建和量测奠定了基础。

Abstract:

This paper proposes an algorithm for trifocal tensor, the method makes use of a set of image point correspondences from three images which taken by RANSAC method accomplish the estimation of the trifocal tensor, which computes briefly and completes easily in the range of the errors. Experimental results indicated that the algorithm proposed is effective and reliable, and it establishes goodwill for camera auto-calibration and 3D-reconstruction.