The ground-based telescope is traditionally tested by autocollimation against a flat mirror. As the aperture of telescope goes larger, the testing can not be performed any more because of the more severe limitation of unstable environment and impracticably flat mirror. The solution of testing wave-front by sub-aperture stitching technique based on Shack-Hartmann wave-front sensor is introduced. This paper begins with a simple description of Sub-Aperture Test (SAT) theory, and explores the methods aiming at more accurate Shack-Hartmann sub-aperture test result. A new trick, that put a plane with holes in the parallel light path, can make sure the location of sub-aperture hit the target. A 32 units Shack-Hartmann wave-front sensor and a 40mm flat mirror is used for testing a optical system 1.8 times larger than the flat mirror and the sub-aperture test results are stitched by two different ways. The experimental results show that error averaging method is superior in error propagation property to stitching one-by-one method, and Peak-to-Valley of the difference between the direct measurement and SAT result is 0.5 wavelength. It is concluded that the technique is useful and has a good application future in practically testing the wave-front error of great telescopes.