知识卡片

普通读书笔记卡 · 8895.c · 生成模型:GPT-6

导数卷积阶数相加

理解等级:高级

连续高斯导数相互卷积后,导数阶数相加,高斯方差也相加。例如沿同一轴的 n、m 阶导数得到 n+m 阶,尺度为 √(σ₁²+σ₂²)。离散采样与有限截断后只是近似,不能期望数值上严格满足所有连续高斯性质。

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 Gaussian derivatives share many of the properties of the Gaussian. The convolution of two Gaussian derivatives of order n and m and variances and result in another Gaussian derivative of order n + m and variance . Proving this property on the spatial domain can be tedious. However, it is trivial to prove it in the Fourier domain.

    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:摘录中关于导数卷积阶数相加的定义、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 Gaussian derivatives share many of the properties of the Gaussian. The convolution of two Gaussian derivatives of order n and m and variances and result in another Gaussian derivative of order n + m and variance . Proving this property on the spatial domain can be tedious. However, it is trivial to prove it in the Fourier domain.