A method for recovering an initial linear state of polarization, of an optical beam after the beam has passed through an optical fiber and whose polarization state was disrupted, attaining power which is directly proportional to the power of the beam before it has passed through the optical fiber. The method comprises: splitting the beam into two beams, a first beam with a first polarization state and a second beam with a second polarization state, the second polarization state being orthogonal to the first polarization state; changing the polarization state of the first beam; and combining the first and second beams in half of their intensities into at least one linearly polarized beam. This way a linearly polarized output beam with power that is directly proportional to the power of the beam prior to its passing through the optical fiber is achieved.