知识卡片
RANSAC先拟合再数共识
反复抽取足够的最小子集,拟合模型,按误差阈值计算全体内点,保留共识集较好的候选,最后用其内点重拟合。直线最小集为两点,单应为四对。阈值定义和误差单位决定谁被计为内点,算法不会自动知道哪些匹配是真实对应。
这张卡片看不懂?
这张卡片已通过人工审查,不再接受失败反馈。
已有反馈 · 0 条
还没有人提交反馈。
在整书笔记中的位置
- · 单应性
原书出处 · 1 处
- · 查看原书依据
The procedure for RANSAC is simple, and the name contains most of the steps of the algorithm. First, randomly select a sufficient set of datapoints to fit the parameters of some model. The could be the parameters that define a line, some other structure, or a homography. Then, compute the model parameters from the randomly sampled set of points. Compute the inliers in the dataset,
参考来源
- 位置:《Foundations of Computer Vision (Antonio Torralba, Phillip Isola etc.)》第41章《单应性》(来源项目:a19c7933-4eda-4176-aebd-30fe7e1ad292;源文件:extracted/Foundations of Computer Vision (Antonio Torralba, Phillip Isola etc.)/_epub-src/OPS/xhtml/c041.xhtml)
- 结论依据:摘录中关于RANSAC先拟合再数共识的定义、比较或条件支撑卡片;原书表述不严谨的范围已在覆盖记录逐项限定,不把案例结果外推为普遍结论。
- 原始内容:The procedure for RANSAC is simple, and the name contains most of the steps of the algorithm. First, randomly select a sufficient set of datapoints to fit the parameters of some model. The could be the parameters that define a line, some other structure, or a homography. Then, compute the model parameters from the randomly sampled set of points. Compute the inliers in the dataset,