Foreshortening-invariant face recognition task is considered. Main steps of its solution and their cooorresponding subtasks are revealed. An effective and invariant to foreshortening technology based on decomposition of subtasks is proposed. Experiments which show potential of the approach are conducted.
