知识卡片

普通读书笔记卡 · 9034.c · 生成模型:GPT-6

纯旋转单应与场景深度无关

理解等级:中级

两相机共享同一投影中心、仅改变朝向时,p2∼K2 R K1^−1p1。射线方向经过旋转后重新投影,因此对应关系不需要场景深度。内参可以不同,但镜头畸变需先处理;手持绕身体转动常有实际平移,近景可能偏离纯旋转假设。

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

      The three images are taken from the same point by rotating the camera. There is no translation between the three camera positions as the photographer took the three images by keeping the camera origin static when moving the camera. Under this condition, the coordinates of the corresponding points across each pair of images are related by a homography independently of how far they are from

    参考来源

    - 位置:《Foundations of Computer Vision (Antonio Torralba, Phillip Isola etc.)》第41章《单应性》(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/c041.xhtml) - 结论依据:摘录中关于纯旋转单应与场景深度无关的定义、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:The three images are taken from the same point by rotating the camera. There is no translation between the three camera positions as the photographer took the three images by keeping the camera origin static when moving the camera. Under this condition, the coordinates of the corresponding points across each pair of images are related by a homography independently of how far they are from