← 返回刚才阅读的位置

知识卡片

普通读书笔记卡 · 8978.d · 生成模型:GPT-6

遮挡轮廓提供深度线索

理解等级:中级

若已确定某边界是遮挡轮廓,它表明可见表面在该处发生前后可见性变化,可帮助推断深度不连续。由图像梯度直接断言遮挡则不成立,因为同样强度边缘还可能来自光照或材质。几何推理需要先说明边缘的物理解释。

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

      A complementary problem to segmentation is boundary detection, because a partitioning of an image can be represented either by partition assignments (segments) or by the boundaries between partitions. Many vision algorithms organize images into different kinds of boundaries between perceptual structures. Early vision algorithms focused on edge detection, where the idea was that the edges in an image should form a good, compact representation of

    参考来源

    - 位置:《Foundations of Computer Vision (Antonio Torralba, Phillip Isola etc.)》第31章《知觉分组》(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/c031.xhtml) - 结论依据:摘录中关于遮挡轮廓提供深度线索的定义、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:A complementary problem to segmentation is boundary detection, because a partitioning of an image can be represented either by partition assignments (segments) or by the boundaries between partitions. Many vision algorithms organize images into different kinds of boundaries between perceptual structures. Early vision algorithms focused on edge detection, where the idea was that the edges in an image should form a good, compact representation of