知识卡片
高斯导数合并两步
在适当可积与可微条件下,∂ₓ(ℓg)=ℓ(∂ₓg)。因此先平滑再求导可合并为一次高斯导数卷积,也可在弱导数意义下处理含阶跃的图像。离散截断核与边界实现可能破坏精确等式,实际比较时须统一这些条件。
这张卡片看不懂?
这张卡片已通过人工审查,不再接受失败反馈。
已有反馈 · 0 条
还没有人提交反馈。
在整书笔记中的位置
- · 图像导数
原书出处 · 1 处
- · 查看原书依据
Let's start with the following observation. For two functions defined in the continuous domain ℓ(x, y) and g(x, y), the convolution and the derivative are commutative: (18.16) This equality is easy to prove in the Fourier domain. If our goal is to compute image derivatives and then blur the output using a differentiable low-pass filter, g(x, y), then instead of computing the derivative of the
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:摘录中关于高斯导数合并两步的定义、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:Let's start with the following observation. For two functions defined in the continuous domain ℓ(x, y) and g(x, y), the convolution and the derivative are commutative: (18.16) This equality is easy to prove in the Fourier domain. If our goal is to compute image derivatives and then blur the output using a differentiable low-pass filter, g(x, y), then instead of computing the derivative of the