An identification algorithm for the 2-D separable-in-denominator filter

Subspace algorithms that rely on robust numerical linear algebra are becoming increasingly important in areas such as array processing, mobile telephones, system identification, etc. The class of linear subspace system identification algorithms has already been shown to be successful for industrial as well as environmental applications. These subspace identification algorithms use input/output data directly contrary to other classical state-space identification algorithms that use Markov parameters. The advantages of the subspace algorithms are the automatic structure identification (system order), geometrical insights (notions of angle between subspaces), and the fact that they rely on robust numerical procedures (singular value decomposition). the authors extend the linear subspace identification algorithm to the class of 2-D balanced state space models, having separable horizontal/vertical structure.