知识卡片

普通读书笔记卡 · 9052 · 生成模型:GPT-6

束调整最小化图像重投影误差

理解等级:中级

Bundle adjustment联合优化相机参数与三维点,使各观测像素接近预测π(K_j[R_j|−R_jT_j]P_i)。残差发生在二维图像空间,原文结尾称“3D重投影误差”容易混淆。式44.2/5为平方欧氏距离和,不是未平方距离和。

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

      and this gives us the final reprojection error equation. One common simplification is to assume that all the cameras have the same intrinsic parameters, which means that K(j) = K. This is the case when the views have been captured by a moving camera with a fixed focal length. The optimization of the reprojection error is also called bundle adjustment.

    参考来源

    - 位置:《Foundations of Computer Vision (Antonio Torralba, Phillip Isola etc.)》第44章《多视图几何与运动恢复结构》(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/c044.xhtml) - 结论依据:摘录中关于束调整最小化图像重投影误差的定义、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:and this gives us the final reprojection error equation. One common simplification is to assume that all the cameras have the same intrinsic parameters, which means that K(j) = K. This is the case when the views have been captured by a moving camera with a fixed focal length. The optimization of the reprojection error is also called bundle adjustment.