When using Lamb waves to detect the damage of the plate structures， the classic minimum power distortion beamforming algorithm can obtain good damage imaging accuracy， but it still can not avoid the influence of dispersion effect on the imaging accuracy. And this algorithm has shortcomings such as complex spectral matrix inversion operation and inability to invert singular spectral matrix， which may reduce computing efficiency significantly. In order to solve the above problems， this paper proposes an improved minimum power distortion beamforming algorithm that is more suitable for Lamb waves detection. The proposed algorithm performs beamforming in the frequency domain to solve the dispersion effect on imaging result， meanwhile combines the least square recursion method and the diagonal loading method to invert the spectral matrix to improve computational efficiency. Experimental and simulation results show that the algorithm can effectively remove the impact of dispersion on the imaging results of damage， thereby effectively solving the problems of artifacts and low imaging resolution in the imaging results of traditional beamforming algorithm. This algorithm also solves the problem of difficulty in inversion of the spectral matrix， which reduces calculation time significantly.