知识卡片

普通读书笔记卡 · 9027 · 生成模型:GPT-6

立体视觉分匹配与三角测量

理解等级:中级

双目深度有两个不同问题:根据相机几何将对应像点转换为三维位置,以及从两幅复杂图像中找到同一场景点的对应。几何关系可缩小匹配搜索范围,却不自动解决遮挡、重复纹理和反射造成的歧义;得到看似合理的对应仍需检验。

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 chapter, we study how to compute depth from a pair of images from spatially offset cameras, such as those of figure 40.1. There are two parts to the depth computation, usually handled separately: (1) analyzing the geometry of the camera projections, which allows for triangulating depth once image offsets are known; and (2) calculating the offset between matching parts of the objects depicted

    参考来源

    - 位置:《Foundations of Computer Vision (Antonio Torralba, Phillip Isola etc.)》第40章《立体视觉》(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/c040.xhtml) - 结论依据:摘录中关于立体视觉分匹配与三角测量的定义、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 chapter, we study how to compute depth from a pair of images from spatially offset cameras, such as those of figure 40.1. There are two parts to the depth computation, usually handled separately: (1) analyzing the geometry of the camera projections, which allows for triangulating depth once image offsets are known; and (2) calculating the offset between matching parts of the objects depicted