← 返回刚才阅读的位置

知识卡片

普通读书笔记卡 · 9075 · 生成模型:GPT-6

空间变流不能直接反向代入

理解等级:高级

对全图常量平移,可以写I₂(x,y)=I₁(x−u,y−v)。一般空间变化流的逆变换却要求解q=p+f(p),不能在目标坐标直接减去原前向流f(q)。本章称两式等价的说明仅应限定在常量平移或正确逆场情形。

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

      where u and v are the element displacement over one unit of time and are also a function of pixel location, u(x, y) and v(x, y), but we will drop that dependency to simplify the notation. The previous relationship is equivalent to saying that ℓ(x, y, t + 1) = ℓ(x – u, y – v, t). To make sure that the reader remains onboard

    参考来源

    - 位置:《Foundations of Computer Vision (Antonio Torralba, Phillip Isola etc.)》第48章《光流估计》(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/c048.xhtml) - 结论依据:摘录中关于空间变流不能直接反向代入的定义、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:where u and v are the element displacement over one unit of time and are also a function of pixel location, u(x, y) and v(x, y), but we will drop that dependency to simplify the notation. The previous relationship is equivalent to saying that ℓ(x, y, t + 1) = ℓ(x – u, y – v, t). To make sure that the reader remains onboard