From a figure you should not do this. The correct procedure is: you use the description of the asymmetric unit, combine it with the space-group symmetry, calculate all fractional coordinates of the unit cell and finally transform the into Cartesian coordinates using the crystal matrix. There are numerous programs which do this, look at CCP14. There are also browser-based software which is able to do this. Search crystal structure presentation software... One is PowderCell which is free available, but there are certainly newer which are more comfortable or more fancy looking. If they are better...I don't know. They look at least better.
Your paper already contains all information you need to know: the structure type, the lattice parameters...i.e. even in case you do not find this phase in a structure data base like the free COD, you can use the coordinates of the structure type and replace the corresponding elements by yours. The lattice parameters tell you the translation lattice so that you can generate your crystal matrix. In so far, this one can even do manually, without any software.
You can download .cif files containing the cell parameters and atomic coordinates for various Perovskite related structures from the AMCSD data base. Here's a link I just got from a Google search I did for "Perovskite" and "AMCSD":