The highly directional laser collimated beam is extensively used as a linear measurement reference in precision measurement. However, random air disturbances like fluctuations in temperature, humidity, and atmospheric pressure can induce jitter, compromising stability and accuracy. This paper introduces a method to counteract these disturbances by using Empirical Mode Decomposition (EMD) combined with Least Squares (LS) fitting for laser collimation signal filtering. The developed signal processing scheme decomposes, reconstructs, and corrects laser signals to remove such interferences. Experiments under various disturbance conditions show that the method proposed results in a standard deviation of fluctuations of 0.01 μm in the horizontal direction and 0.02 μm in the vertical direction after processing the laser alignment signal following laser drift compensation, significantly improves the stability of laser collimation measurements affected by atmospheric randomness.