Abstract: Atmospheric turbulence is an important form of movement in the atmosphere, which makes the vertical and horizontal's exchange interaction of momentum, heat, water vapor and pollutants significantly enhanced, and this interference has a great impact on the target resolution of optical imaging system. There also have lucky regions in the degraded images because of turbulence, so the appropriate algorithm can obtain high resolution restored image. To obtain the atmospheric turbulence blurred images which contains lucky regions, the artificial turbulence was used in the laboratory and combined with the short-exposure technique to take a serial of atmospheric turbulence blurred images. The rectangle overlapped partition method was used and the image restoration algorithm was improved based on nonlinear partial-derivative equations(PDE's), to process the obtained short exposure serial images. The results show that, the image quality of the composite image is improved obviously, this algorithm has great restoration effect on images' quality degradation caused by atmospheric turbulence.