Similar to the dataset parameters, camera parameters are stored in a JSON format used by all the tools. Each camera is defined by a name, an extrinsic matrix and an intrinsic matrix.
The extrinsic matrix Rt
is a 4x4 rigid transformation matrix in homogeneous coordinates. The last row is always
0 0 0 1
. For a point w = (X,Y,Z,1)
in world space, v = Rt * w
results in the same point
in view space of this camera.
The tool camera/export_mpeg exports the JSON file to a MPEG-format camera file usable with VSRS and DERS, and automatically does the necessary conversion of the extrinsic matrices. camera/import_mpeg imports a MPEG-format camera file into a JSON file, and does the inverse conversion.