← 返回起始笔记位置

知识卡片

普通读书笔记卡 · 8915 · 生成模型:GPT-6

窗口使频率分析局部化

理解等级:中级

全局傅里叶系数把整幅图像的信息汇总,单个系数不能直接定位结构。用高斯窗乘正弦或复指数得到 Gabor 核,使每个响应关注附近的特定方向与频率。高斯仍有无限支撑,这里的“局部”指快速衰减的有效范围,不等于严格有限窗口。

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

      Figure 22.1(a) shows a one-dimensional (1D) sine function. This function has infinite support. When used as a representation, the Fourier coefficients are obtained as the scalar product between the sine wave and the input image. Each Fourier coefficient involves a point-wise multiplication between the wave and the input image and then the sum of the result. As a consequence of the infinite support of the

    参考来源

    - 位置:《Foundations of Computer Vision (Antonio Torralba, Phillip Isola etc.)》第22章《滤波器组》(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/c022.xhtml) - 结论依据:摘录中关于窗口使频率分析局部化的定义、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:Figure 22.1(a) shows a one-dimensional (1D) sine function. This function has infinite support. When used as a representation, the Fourier coefficients are obtained as the scalar product between the sine wave and the input image. Each Fourier coefficient involves a point-wise multiplication between the wave and the input image and then the sum of the result. As a consequence of the infinite support of the