The work aims to study the roughness parameters variation of morphology with different height distribution functions (HDF) and autocorrelation functions (ACF). The three-dimensional random surfaces of specific roughness parameters (eg., gradient, peak value, steepest descent autocorrelation length, texture aspect ratio and height variance) were constructed by using the two-dimensional digital filter method. The parameters of each morphology were compared and analyzed. Curvature of arithmetic mean peak did not depend on HDF and ACF; root-mean-square gradient, developed interfacial area ratio and peak density were heavily influenced by HDF and ACF. The arithmetic mean peak curvature can not be used to characterize the HDF and ACF. When surfaces generated from the same HDF were analyzed, texture aspect ratio, steepest descent autocorrelation length, root-mean-square gradient, developed interfacial area ratio and peak density should be compared. When surfaces generated from the different HDFs were analyzed, height parameters such as gradient, peak value, height root-mean-square, maximum height, maximum valley value and maximum peak height should be compared.