← 返回刚才阅读的位置

知识卡片

普通读书笔记卡 · 8959.b · 生成模型:GPT-6

保留样本不保证学到分布

理解等级:中级

非参数纹理方法把参考图本身作为表示,保留样本像素,避免先压成少量统计。但样本有限、邻域选择和检索准则仍会遗漏未知变化,不能因此称其完整掌握真实纹理分布。保存数据的无损与生成模型的充分性是不同问题。

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

      In this nonparametric model, the representation of a texture is the reference image itself. There is not a low-dimensional representation as it is the case of parametric texture models. The advantage is that no information is lost in this representation.

    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:摘录中关于保留样本不保证学到分布的定义、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:In this nonparametric model, the representation of a texture is the reference image itself. There is not a low-dimensional representation as it is the case of parametric texture models. The advantage is that no information is lost in this representation.