知识卡片

普通读书笔记卡 · 9043.b · 生成模型:GPT-6

消失点内参估计先检查正定

理解等级:高级

把v_i^T W v_j=0堆成Aw=0,单位范数下取最小奇异值右奇异向量,再还原对称W。W应为正定且满足设定结构;噪声或错误方向可能给非物理解。这里得到的是最小奇异向量,而非未定义的“最小特征值”直接SVD输出。

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 linear equation can be solved with the singular value decomposition (SVD) and the result is the eigenvector with the smallest eigenvalue. Once we have W, we can find K by using the Cholesky factorization, which decomposes a matrix as the product of an upper triangular matrix and its transpose. The last step is to normalize the result by normalizing all the matrix values so

    参考来源

    - 位置:《Foundations of Computer Vision (Antonio Torralba, Phillip Isola etc.)》第42章《单视图测量》(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/c042.xhtml) - 结论依据:摘录中关于消失点内参估计先检查正定的定义、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 linear equation can be solved with the singular value decomposition (SVD) and the result is the eigenvector with the smallest eigenvalue. Once we have W, we can find K by using the Cholesky factorization, which decomposes a matrix as the product of an upper triangular matrix and its transpose. The last step is to normalize the result by normalizing all the matrix values so