← 返回刚才阅读的位置

知识卡片

普通读书笔记卡 · 9028.c · 生成模型:GPT-6

远处深度对视差误差更敏感

理解等级:高级

由Z=f_x B/δ,一阶误差幅度约|ΔZ|≈f_x B|Δδ|/δ²=Z²|Δδ|/(f_x B)。相同像素视差误差在远处形成更大深度误差,因此视差空间噪声常更便于建模。该近似针对小误差,不说明真实误差必然独立同分布。

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

      One important question is how we want to represent the output. We are ultimately interested in estimating the 3D structure of the scene. We can estimate the disparity or the depth. Disparity might be easier to estimate as it is independent of the camera parameters. It is also easier to model noise, as we can assume that noise is independent of disparity. However, when using

    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:摘录中关于远处深度对视差误差更敏感的定义、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:One important question is how we want to represent the output. We are ultimately interested in estimating the 3D structure of the scene. We can estimate the disparity or the depth. Disparity might be easier to estimate as it is independent of the camera parameters. It is also easier to model noise, as we can assume that noise is independent of disparity. However, when using