← 返回起始笔记位置

知识卡片

普通读书笔记卡 · 9054.g · 生成模型:GPT-6

稠密光度优化引入外观假设

理解等级:中级

稠密方法可利用大部分像素的光度差获得丰富结构,但对应亮度稳定、足够可见重叠和合理运动等假设会限制使用。朗伯近似仅减少视向相关反射,不自动保证照明和曝光不变。稀疏与稠密的快慢、精度不能脱离分辨率、算法与硬件一概排名。

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 dense SFM, the 3D reconstruction is computed for every (or most) of the image pixels. The resulting 3D reconstructions are rich and visually informative, and are useful in applications like robot navigation or mixed reality. Dense SFM approaches usually optimize photometric error, that is, the difference in grayscale or red-green-blue (RGB) intensities of corresponding pixels across images. This is in contrast to the point

    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:摘录中关于稠密光度优化引入外观假设的定义、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 dense SFM, the 3D reconstruction is computed for every (or most) of the image pixels. The resulting 3D reconstructions are rich and visually informative, and are useful in applications like robot navigation or mixed reality. Dense SFM approaches usually optimize photometric error, that is, the difference in grayscale or red-green-blue (RGB) intensities of corresponding pixels across images. This is in contrast to the point