知识卡片

普通读书笔记卡 · 9036.a · 生成模型:GPT-6

最小子集必须避免退化

理解等级:中级

单应RANSAC每轮应取不同的四对点,并排除共线、重复或无法拟合可逆H的配置。即使样本全是内点,也可能因退化或噪声而没有得到好模型。因此迭代成功概率的简化推导不是无条件正确模型保证。

这张卡片看不懂?

已有反馈 · 0 条

    还没有人提交反馈。

    在整书笔记中的位置

    原书出处 · 1 处

    • · 查看原书依据

      In the case of the homography estimation, we need eight equations to use the DLT algorithm to estimate H. As each point contributes two equations, we need four corresponding points. We can sample multiple randomly selected sets of four points and use RANSAC to estimate the homography between two images.

    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:摘录中关于最小子集必须避免退化的定义、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:In the case of the homography estimation, we need eight equations to use the DLT algorithm to estimate H. As each point contributes two equations, we need four corresponding points. We can sample multiple randomly selected sets of four points and use RANSAC to estimate the homography between two images.