知识卡片

普通读书笔记卡 · 8957.a · 生成模型:GPT-6

子带直方图压缩纹理信息

理解等级:中级

Heeger–Bergen方法用多个方向和尺度的可转向金字塔分解图像,保存各子带、低通残差及原图的直方图。书中三尺度六方向例子含18个方向子带。这些分布保留响应大小出现频率,却不直接记录响应在何处或各子带如何共同出现。

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 28.7 shows the main idea behind the approach proposed by Heeger and Bergen to implement the analysis and synthesis functions. First, the reference texture is decomposed multiple orientation and scales as the outputs of many oriented filters over different spatial scales. The transform should represent all spatial frequencies of the image, so that all aspects of a texture can be synthesized, and the subbands

    参考来源

    - 位置:《Foundations of Computer Vision (Antonio Torralba, Phillip Isola etc.)》第28章《纹理》(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/c028.xhtml) - 结论依据:摘录中关于子带直方图压缩纹理信息的定义、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 28.7 shows the main idea behind the approach proposed by Heeger and Bergen to implement the analysis and synthesis functions. First, the reference texture is decomposed multiple orientation and scales as the outputs of many oriented filters over different spatial scales. The transform should represent all spatial frequencies of the image, so that all aspects of a texture can be synthesized, and the subbands