A pupil splitting lens that splits a pupil of a light beam from a deflector into a plurality of light beams in a main scanning direction of the deflector and a pupil splitting lens that splits the light beam into a plurality of light beams in a sub-scanning direction are disposed relative to a light detection element. The light detection element detects image forming positions of the four light beams formed as a result of the splitting by the plurality of pupil splitting lenses. A CPU determines the amount of focus shift and the amount of image forming position shift from these four image forming positions.