知识卡片

普通读书笔记卡 · 8891.a · 生成模型:GPT-6

二项核由重复两抽头卷积构成

理解等级:中级

将[1,1]重复卷积n次,得到Pascal三角第n行的二项系数,长度n+1、系数和2ⁿ。除以2ⁿ后可看成归一化离散核,其中心化空间方差为n/4。参数n是阶数,并非核长度或σ本身。

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 practice, there are very efficient discrete approximations to the Gaussian filter that, for certain σ values, have nicer properties than when working with discretized Gaussians. One common approximation of the Gaussian filter is to use binomial coefficients [77]. Binomial filters are obtained by successive convolutions of the box filter [1, 1].

    参考来源

    - 位置:《Foundations of Computer Vision (Antonio Torralba, Phillip Isola etc.)》第17章《模糊滤波器》(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/c017b.xhtml) - 结论依据:摘录中关于二项核由重复两抽头卷积构成的定义、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 practice, there are very efficient discrete approximations to the Gaussian filter that, for certain σ values, have nicer properties than when working with discretized Gaussians. One common approximation of the Gaussian filter is to use binomial coefficients [77]. Binomial filters are obtained by successive convolutions of the box filter [1, 1].