The inversion algorithm of differential image collumn motion(DCIM) lidar measuring turbulence profile was proposed based on generalized Hufnagel-Valley model. The smoothing function of data was derived. Through numerical simulation of HV 5/7 model, some results were obtained. Without random error, the maximum difference between the inversion profile and HV 5/7 model profile is about 0.00order of magnitude. The absolute relative errors of r0 and the whole layer θ calculated by the inversion profiles are both less than 0.5%. With 10% random errors, the maximum difference between the inversion profiles and HV 5/7 model profile is about 0.5 order of magnitude below 20 km while it is bigger than 20 km. The absolute relative errors of r0 and the whole layer θ computed by the inversion profile are less than 10% and 5%. So it meets the actual application of laser propagation in the atmosphere. The smoothing function is suitable for practical application because it does not affect the distribution of the inversion profile with height. Finally, the universality of the algorithm was verified and the algorithm can be used to DCIM lidar to get profile.