← 返回刚才阅读的位置

知识卡片

普通读书笔记卡 · 8894 · 生成模型:GPT-6

梯度域编辑需重建

理解等级:中级

把图像编码为横纵导数后,可以在指定掩膜内修改梯度,再通过重建获得像素图像。书中消除 STOP 字样的例子将文字梯度置零,使周围近似平坦的红色区域延续进去。它依赖局部背景与边界约束,不能保证任意纹理都被合理补全。

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

      One of the interests of transforming an image into a different representation than pixels is that it allows us to manipulate aspects of the image that would be hard to control using the pixels directly. Figure 18.5 shows an example of image editing using derivatives.

    参考来源

    - 位置:《Foundations of Computer Vision (Antonio Torralba, Phillip Isola etc.)》第18章《图像导数》(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/c018.xhtml) - 结论依据:摘录中关于梯度域编辑需重建的定义、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:One of the interests of transforming an image into a different representation than pixels is that it allows us to manipulate aspects of the image that would be hard to control using the pixels directly. Figure 18.5 shows an example of image editing using derivatives.