← 返回刚才阅读的位置

知识卡片

普通读书笔记卡 · 8875.b · 生成模型:GPT-6

一般线性滤波允许位置专属权重

理解等级:中级

一般离散线性系统写ℓout[n]=Σₖh[n,k]ℓin[k],权重依赖输出位置和输入位置。矩阵H的每行规定一个输出如何组合输入,不要求只看局部,也不要求每行使用同样模式。卷积是在此基础上再加结构约束。

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

      where each output value ℓout [n] is a linear combination of values of the input signal ℓin [n] with weights h [n, k]. The value h [n, k] is the weight applied to the input sample ℓin [k] to compute the output sample ℓout [n]. To help visualizing the operation performed by the linear filter it is useful to write it in matrix form:

    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:摘录中关于一般线性滤波允许位置专属权重的定义、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:where each output value ℓout [n] is a linear combination of values of the input signal ℓin [n] with weights h [n, k]. The value h [n, k] is the weight applied to the input sample ℓin [k] to compute the output sample ℓout [n]. To help visualizing the operation performed by the linear filter it is useful to write it in matrix form: