This contribution deals with the identification of gloss as a function of roughness using particle swarm optimization (PSO) methods. The proposed PSO methods use a least squares method (LSM) as a cost function to be optimized. The nonparametric identification structure uses a Gaussian regressor characterized by three parameters to be estimated. Three different algorithms are proposed: a global classical PSO, an intertwined PSO structure and a PSO structure combined with a linear regression method obtained using a logarithmical transformation. Results using measured data are shown at the end of this analysis to compare the three different techniques.