The invention provides an optimal sampling interval condition based roughness coefficient fractal evaluation model constructing method. The method includes steps of 1), selecting a high pixel image ofm standard outlines of Barton, extracting data information of the m standard section lines in the image and acquiring outlines of the m standard section lines and performing post processing; 2) programming in matlab according to a yardstick method, importing the information of the contour lines of a structural face in the step 1), setting different sampling intervals r, obtaining N(r) values corresponding to the m standard contour lines and calculating out a fractal dimension D; 3) performing analysis on dispersion degree of the fractal dimension D in 91 sampling interval sections; 4), exploring the change rule of the fractal dimension D in different sampling interval gaps, obtaining the optimal sampling interval r, and constructing a function model of a roughness coefficient of the structural face in the optimal sampling interval. According to the invention, the structural face roughness coefficient JRC can be estimated more accurately according to the optimal sampling interval.