We assemble a database of 15 HSB and 15 LSB galaxies, for which surface brightness density and spectroscopic rotation curve data are both available and representative for various morphologies. We use this dataset to test the Navarro-Frenk-White, the Einasto, and the pseudo-isothermal sphere dark matter (DM) models. We investigate the compatibility of the pure baryonic model and baryonic plus one of the three DM models with observations on the assembled galaxy database. When a DM component improves the fit with the spectroscopic rotational curve, we rank the models according to the goodness of fit to the datasets. We constructed the spatial luminosity density of the baryonic component based on the surface brightness profile of the galaxies. We estimated the mass-to-light (M/L) ratio of the stellar component through a previously proposed color-mass-to-light ratio relation (CMLR). We assumed an axissymetric baryonic mass model with variable axis ratios together with one of the three DM models to provide the theoretical rotational velocity curves, and we compared them with the dataset. In a second attempt, we addressed the question whether the dark component could be replaced by a pure baryonic model with fitted M/L ratios. We employed the Akaike information criterion (AIC) to establish the performance of the best-fit models. For 7 galaxies, neither model fits the dataset within the 1{\sigma} confidence level. For the other 23 cases, one of the models with DM explains the rotation curve data best. According to the AIC, the PSE emerges as most favored in 14 cases, followed by the NFW (6 cases) and the Einasto (3 cases) DM models. We find that the pure baryonic model with fitted M/L ratios falls within the 1{\sigma} confidence level for 10 HSB and 2 LSB galaxies, at the price of growing the M/Ls on average by a factor of two, but the fits are inferior compared to the best-fitting DM model.