Full-waveform airborne LiDAR （FWL） is able to record complete echo signals as waveforms， including useful information such as elevation details and backscatter coefficients of the target， but the waveform information data cannot be obtained directly. Waveform decomposition is an important method to process waveform data to extract effective information. In view of the shortcoming of common used parameter optimization algorithm in waveform decomposition which is sensitive to initial value and prone to local optimization， a waveform decomposition method based on Modified Differential Evolution （MDE） algorithm is proposed： the generalized Gaussian function is taken as the model， after the initial estimation， a global MDE optimization algorithm is used for the parameter optimization， and the point cloud is finally generated. Experimental results show that， compared with the waveform decomposition method based on other optimization algorithms， this method has been obviously improved in terms of the decomposition and point position accuracy.