知识卡片

普通读书笔记卡 · 8910.d · 生成模型:GPT-6

理想截止带来全局核

理解等级:中级

有限DFT中的矩形低通对应周期的离散 sinc 型核,通常支撑覆盖整幅图像。它能按选定频带严格抑制分量,但直接空间卷积成本高,并对边界与频带端点处理敏感。所谓“理想”描述频率选择,不代表空间图像没有振铃或视觉伪影。

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

      The ideal anti-aliasing filter is a box in the Fourier domain with a cut-off frequency of L = N/(2k) (using the notation introduced in chapter 20). The inverse Fourier transform of a box in the frequency domain, with width 2L + 1 = N/k + 1, is the discrete sinc function of length N in the spatial domain:

    参考来源

    - 位置:《Foundations of Computer Vision (Antonio Torralba, Phillip Isola etc.)》第21章《图像下采样与上采样》(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/c021.xhtml) - 结论依据:摘录中关于理想截止带来全局核的定义、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 ideal anti-aliasing filter is a box in the Fourier domain with a cut-off frequency of L = N/(2k) (using the notation introduced in chapter 20). The inverse Fourier transform of a box in the frequency domain, with width 2L + 1 = N/k + 1, is the discrete sinc function of length N in the spatial domain: