For free-space optical communication links, the light spot collected by the photodetector at the receiving terminal is not an ideal light spot that is affected by atmospheric turbulence. The light spot collected by the photodetector will also be accompanied by various noises. More importantly, the presence of all noise will bring errors to acquire the light spot's center. As a result, the tracking error can affect the stability of the optoelectronic tracking system. Therefore, it is necessary to remove noise from the collected images. The method of removing noise needs to be effective, but it cannot bring a large amount of calculation to affect the real-time performance. The calculation amount of wavelet transform is small, and the effect of noise removal is better, which can focus on local details with arbitrary expansion coefficients. An improved wavelet denoising method is proposed. The long-distance verification experiment (11.16 km) verified the effectiveness of this approach, compared with the traditional method. Furthermore, to the best of our knowledge, this new approach would be beneficial for the design of optical communication systems.