基于高精度匹配点的对极几何估计

提出了一种以Harris算子和空间位置约束条件自动提取的初始特征点为引导,利用小面元模型进行灰度分布的拟合曲面求极值点的超精度特征点定位方法。并利用超精度的特征点估计立体图像对的对极几何约束关系。通过对真实图像的对比实验证明该方法能有效的提高基本矩阵估计精度,与传统的基于像素级特征点的估计算法相比,平均余差值降低了3.4%~21.4%,算法运行速度快,能在视觉测量中有效地提高测量精度。

Epipolar Geometry Estimation Using High Precision Feature Points

A new algorithm for high precision feature point detection is proposed in this paper;it can be gained from the extremal points by fitting a facet surface to the Harris pre-detected feature points.The experiment results of estimation epipolar geometry using the high precision feature points in real images show that it can improve the accuracy of the fundamental matrix,Compared with the estimation method using the pixel-level feature points,the average residual errors of the fundamental matrix are reduced 3.4%~21.4%,the algorithm is fast in computation and also it can improve the accuracy of the vision measurement system.