The work aims to study the effects of autocorrelation length on reconstructed surface characterization parameters in the process of computer simulation, and lay a foundation for exploring the relationship among different surface characterization parameters, and also the relationship between autocorrelation function (ACF) and surface properties of usability. Computer simulation surface test of 3D rough surface morphology was carried out 16 times for various combinations of autocorrelation length 3, 10, 30, 100 in two directions using the same specific parameters (RMS, Skewness, Kurtosis) in the new FFT method proposed by the author. The effects of parameters which were correlated with morphological features were analyzed from the perspective of combinations of autocorrelation length in different direction. Mean summit curvature, summit density and RMS slope were negatively correlated with autocorrelation length in two different directions. However, the fastest descending autocorrelation length showed a positive correlation, and texture aspect ratio was related to proximity of autocorrelation length in two directions. Summit density and the fastest descending auto-correlation length were both sensitive to smaller autocorrelation length among different autocorrelation lengths, but less sensitive to changes in larger autocorrelation length. Mean summit curvature and RMS slope showed the same high sensitivity. Height distribution parameters among surface characterization parameters are not correlated with autocorrelation length on computer simulation surface, while structure proportion parameters are significantly correlated with autocorrelation length.