来源wiki pedia: http://en.wikipedia.org/wiki/Bundle_adjustment
光束平差法的最终目的归结为:减少观测图像的点和参考图像(预测图像)的点之间位置投影变换(再投影)误差。这最小化误差算法使用的是最小二乘算法,目前
使用最为成功是Levenberg-Marquardt, 它具有易于实现,对大范围的初始估计能够快速收敛的优点。
Bundle adjustment amounts to jointly refining a set of initial camera and structure parameter estimates for finding the set of parameters that most accurately predict the locations of the observed points in the set of available images. More formally, assume that 3D points are seen in
views and let
be the projection of the
th point on image
. Let
denote the binary variables that equal 1 if point
is visible in image
and 0 otherwise. Assume also that each camera
is parameterized by a vector
and each 3D point
by a vector
. Bundle adjustment minimizes the total reprojection error with respect to all 3D point and camera parameters, specifically
where is the predicted projection of point
on image
and
denotes the Euclidean distance between the image points represented by vectors
and
. Clearly, bundle adjustment is by definition tolerant to missing image projections and minimizes a physically meaningful criterion.
Software
- sba: A Generic Sparse Bundle Adjustment C/C++ Package Based on the Levenberg–Marquardt Algorithm (C, Matlab)
- ssba: Simple Sparse Bundle Adjustment package based on the Levenberg–Marquardt Algorithm (C) with LGPL license.
- OpenCv: Computer Vision library in the contrib module.
- mcba: Multi-Core Bundle Adjustment (CPU/GPU).
- libdogleg: General-purpose sparse non-linear least squares solver, based on Powell's dogleg method. LGPL.
- ceres-solver: A Nonlinear Least Squares Minimizer with BSD license