知识卡片

普通读书笔记卡 · 8959.c · 生成模型:GPT-6

邻域大小决定结构视野

理解等级:中级

上下文太小可能只见到圆弧,无法要求生成完整圆;扩大邻域才可能同时约束元素形状及其间距。但更大邻域也减少有效匹配、提高搜索成本并容易重复参考图内容。它应覆盖目标结构,而非无条件越大越好。

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 28.11 shows how the synthesized texture from the Efros-Leung algorithm [115] changes as a function of size of the context region. Figure 28.11(a) shows the source texture; figure 28.11(b) shows how a very small context region (the grid of figure 28.10) only covers a part of each circle and the synthesized texture doesn't show complete circles; figure 28.11(c) shows how a larger context region

    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:摘录中关于邻域大小决定结构视野的定义、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 28.11 shows how the synthesized texture from the Efros-Leung algorithm [115] changes as a function of size of the context region. Figure 28.11(a) shows the source texture; figure 28.11(b) shows how a very small context region (the grid of figure 28.10) only covers a part of each circle and the synthesized texture doesn't show complete circles; figure 28.11(c) shows how a larger context region