知识卡片
重投影损失回到图像空间
给定3D点与观测像素,计算预测 p̂_i=π(K[R|−RT]P_i),最小化 Σ||p_i−p̂_i||²。π包含去齐次操作,也可纳入畸变模型。误差应报告像素单位、聚合方式和分布;平方误差、RMSE与平均欧氏距离并非同一指标。
这张卡片看不懂?
这张卡片已通过人工审查,不再接受失败反馈。
已有反馈 · 0 条
还没有人提交反馈。
在整书笔记中的位置
- · 相机建模与标定
原书出处 · 1 处
- · 查看原书依据
We are given 3D points, Pi, and their corresponding 2D projections, pi. We start with some estimated camera parameters K, R, and T. We can use these parameters to project the 3D points and get estimated 2D locations, , of their projections. The reprojection error is the distance between those points, as shown in figure 39.14.
参考来源
- 位置:《Foundations of Computer Vision (Antonio Torralba, Phillip Isola etc.)》第39章《相机建模与标定》(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/c039.xhtml)
- 结论依据:摘录中关于重投影损失回到图像空间的定义、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:We are given 3D points, Pi, and their corresponding 2D projections, pi. We start with some estimated camera parameters K, R, and T. We can use these parameters to project the 3D points and get estimated 2D locations, , of their projections. The reprojection error is the distance between those points, as shown in figure 39.14.