← 返回刚才阅读的位置

知识卡片

普通读书笔记卡 · 8875.c · 生成模型:GPT-6

平移等变使输出随输入同移

理解等级:中级

若输入平移后,输出也以相同位移平移,系统具有平移等变性。本章传统LTI用“平移不变系统”称此性质,但并非输出保持原样的不变性。有限图像的裁剪、填充和步幅会影响具体成立范围,不能只凭卷积名称断言所有边界都严格等变。

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

      for any n0, m0. This property is called equivariant with respect to translation. Translation invariance and linearity impose a strong constraint on the form of equation (15.7). In that case, the linear system becomes a convolution of the image data with some convolution kernel. This operation is very important and we devote the rest of this chapter to study its properties. 15.4 Convolution The convolution,

    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:摘录中关于平移等变使输出随输入同移的定义、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:for any n0, m0. This property is called equivariant with respect to translation. Translation invariance and linearity impose a strong constraint on the form of equation (15.7). In that case, the linear system becomes a convolution of the image data with some convolution kernel. This operation is very important and we devote the rest of this chapter to study its properties. 15.4 Convolution The convolution,