Stereo matching is the process of establishing correspondence between different perspective images of the same scene. A global optimal method that can deliver a dense disparity map is prefer able to a method producing sparse displacement results or a method based on local optimization. We present an effective global optimized stereo matching approach that produces a dense displacement map and an occlusion map. The global matching cost and various constraints, including matching uniqueness and ordering, and local smoothness along and across epipolar lines, are all cast into a novelly configured maximum flow graph. The correspondence between the associated minimum cut and the defined stereo problem guarantees a global optimized disparity solution confined by those geometric constraints while still preserving discontinuities. Test results show the efficacy of the algorithm.