In this paper, to manage the lidar signal noise, an effective denoising method of wavelet packet transform decomposition and reconstruction based on data segmentation is proposed. This can provide high-quality data for extinction coefficient retrieval. Herein, to verify the proposed denoising method feasibility, the differences between the original radar data, the extinction coefficient retrieved after wavelet packet processing, and the extinction coefficient retrieved after segmented wavelet packet processing were compared. In the numerical simulation using Bump and Block test signals, the value of mean square error (MSE) and R of the wavelet packet algorithm based on data segmentation was found to be better than that of the other wavelet packet algorithms. The corresponding experimental result indicates that the segmented wavelet packet denoising method can improve the results of the next-level inversion extinction coefficient via denoising, effectively protect the detailed structure of the low-altitude lidar extinction profile, and suppress the noise at high altitudes. Furthermore, the changing of the extinction coefficient pseudo-color image of air aerosol with time and space shows the feasibility and usefulness of the proposed method.