← 返回刚才阅读的位置

知识卡片

普通读书笔记卡 · 8923.b · 生成模型:GPT-6

镜像滤波压缩成独立系数

理解等级:中级

本章Haar式QMF用归一化的相邻和与差,分别表示低频与高频,再以步长2抽取,使总系数数目等于原样本数。正交归一化与对应合成保证重建;若省略1/√2或随意更改通道增益,转置合成就不再直接等于原信号。

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 quadrature mirror filter (QMF) transform is an example of self-inverting transform [10]. For 1D signals of length 4, the QMF transform can be written as: This is equivalent to the convolution of the input signal with two orthogonal kernels, [1, 1] and [1, –1], with a stride of 2. This transform also has a multiscale version that is also self-inverting: You can check that

    参考来源

    - 位置:《Foundations of Computer Vision (Antonio Torralba, Phillip Isola etc.)》第23章《图像金字塔》(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/c023.xhtml) - 结论依据:摘录中关于镜像滤波压缩成独立系数的定义、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 quadrature mirror filter (QMF) transform is an example of self-inverting transform [10]. For 1D signals of length 4, the QMF transform can be written as: This is equivalent to the convolution of the input signal with two orthogonal kernels, [1, 1] and [1, –1], with a stride of 2. This transform also has a multiscale version that is also self-inverting: You can check that