← 返回起始笔记位置

知识卡片

普通读书笔记卡 · 9029.b · 生成模型:GPT-6

角点在各方向都易定位

理解等级:中级

对小平移Δ,局部强度平方差可一阶近似为E(Δ)≈Δ^T MΔ,其中M=Σ_P [I_x,I_y]^T[I_x,I_y]。若最小特征值也大,则任何单位方向的小位移都引起较大变化,位置容易确定;单条边只有一个强变化方向,平坦区两个方向都弱。

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

      Good image features are positions in the image that are easy to localize, given only the image intensities. The Harris corner detector [186] identifies image points that are easy to localize by evaluating how the sum of squared intensity differences over an image patch change under a small translation of the image. If the squared intensity changes quickly with patch translation in every direction, then

    参考来源

    - 位置:《Foundations of Computer Vision (Antonio Torralba, Phillip Isola etc.)》第40章《立体视觉》(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/c040.xhtml) - 结论依据:摘录中关于角点在各方向都易定位的定义、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:Good image features are positions in the image that are easy to localize, given only the image intensities. The Harris corner detector [186] identifies image points that are easy to localize by evaluating how the sum of squared intensity differences over an image patch change under a small translation of the image. If the squared intensity changes quickly with patch translation in every direction, then