In class, a question was raised regarding QR decomposition. The standard QR decomposition takes a matrix $M$ as input, and decomposes it into the product two matrices: an orthogonal matrix $Q$, and an upper triangular matrix $R$, such that $M = QR$. However, in our case, we actually want a slightly different decomposition where $M = RQ$.
Fortunately, there are multiple analogous forms of the QR decomposition: $QR$, $QL$, $RQ$, and $LQ$, where $A$ is a orthogonal matrix, $R$ is an upper triangular matrix, and $L$ is a lower triangular matrix. For recovering the intrinsics and rotation, we would use the $RQ$ decomposition.
In class, a question was raised regarding QR decomposition. The standard QR decomposition takes a matrix $M$ as input, and decomposes it into the product two matrices: an orthogonal matrix $Q$, and an upper triangular matrix $R$, such that $M = QR$. However, in our case, we actually want a slightly different decomposition where $M = RQ$.
Fortunately, there are multiple analogous forms of the QR decomposition: $QR$, $QL$, $RQ$, and $LQ$, where $A$ is a orthogonal matrix, $R$ is an upper triangular matrix, and $L$ is a lower triangular matrix. For recovering the intrinsics and rotation, we would use the $RQ$ decomposition.