← 返回起始笔记位置

知识卡片

普通读书笔记卡 · 8952.a · 生成模型:GPT-6

颜色去相关不等于独立

理解等级:中级

对颜色通道作PCA可以移除线性二阶相关,再分别处理各主成分并逆变换回原颜色空间。但去相关只在联合高斯等条件下推出独立。单独随机化各通道也不能保证完全保留原RGB联合分布,因此应把它看作近似合成策略。

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

      We can also sample new images by matching the parameter of individual natural images. Figure 27.12 shows four examples. To process color images, we first look for a color space that decorrelates color information for each image. This can be done by applying principal component analysis (PCA) to the three color channels. This results in a 3×3 matrix that will produce three new channels that

    参考来源

    - 位置:《Foundations of Computer Vision (Antonio Torralba, Phillip Isola etc.)》第27章《统计图像模型》(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/c027b.xhtml) - 结论依据:摘录中关于颜色去相关不等于独立的定义、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:We can also sample new images by matching the parameter of individual natural images. Figure 27.12 shows four examples. To process color images, we first look for a color space that decorrelates color information for each image. This can be done by applying principal component analysis (PCA) to the three color channels. This results in a 3×3 matrix that will produce three new channels that