← 返回起始笔记位置

知识卡片

结构图卡 · 8879.f · 生成模型:GPT-6

线性滤波的建模判断链

理解等级:中级

从一般系统逐步增加线性、同位移等变与局部支撑约束,得到可用卷积核描述的滤波。有限数组实现还须明确边界和输出裁剪。

结构图:

flowchart TD
 A[信号到信号的系统] --> B{满足线性叠加}
 B -->|是| C[一般权重矩阵H]
 C --> D{同位移等变}
 D -->|是| E[权重仅依相对偏移]
 E --> F[卷积核描述]
 F --> G[声明核原点与是否反转]
 G --> H[边界扩展及输出裁剪]
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

      Linear systems are still too general for us. So let's consider an even smaller family of systems: linear translation invariant (LTI) systems. Translation invariant systems are motivated by the following observation: typically, we don't know where within the image we expect to find any given item (figure 15.6), so we often want to process the image in a spatially invariant manner, the same processing algorithm

    参考来源

    - 位置:《Foundations of Computer Vision (Antonio Torralba, Phillip Isola etc.)》第15章《线性图像滤波》(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/c015b.xhtml) - 结论依据:摘录中关于线性滤波的建模判断链的定义、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:Linear systems are still too general for us. So let's consider an even smaller family of systems: linear translation invariant (LTI) systems. Translation invariant systems are motivated by the following observation: typically, we don't know where within the image we expect to find any given item (figure 15.6), so we often want to process the image in a spatially invariant manner, the same processing algorithm