← 返回刚才阅读的位置

知识卡片

普通读书笔记卡 · 8892.d · 生成模型:GPT-6

差分不是理想高通

理解等级:高级

连续求导的频域乘子为 jω;两点差分的乘子是 e^(−jω/2)·2j sin(ω/2),中心差分是 j sinω。二者只在低频附近近似理想斜率,且相位与高频响应不同。比较导数核需要同时检查幅度、相位和归一化,不能只看其是否增强高频。

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 continuous Fourier domain, derivation is equivalent to multiplying by jw. In the discrete domain, the DFT of a signal derivative will depend on the approximation used. Let's study now the DFT of the two approximations that we have discussed here, d0 and d1.

    参考来源

    - 位置:《Foundations of Computer Vision (Antonio Torralba, Phillip Isola etc.)》第18章《图像导数》(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/c018.xhtml) - 结论依据:摘录中关于差分不是理想高通的定义、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 continuous Fourier domain, derivation is equivalent to multiplying by jw. In the discrete domain, the DFT of a signal derivative will depend on the approximation used. Let's study now the DFT of the two approximations that we have discussed here, d0 and d1.