Aiming at the problem that the physical model has poor antinoise ability and is easy to overfit, a PROSAIL model was proposed by combining VMG (VARI (visible atmospherically resistant index), MGRVI (modified green red vegetation index) and GRRI (green red ratio index)) to retrieve the leaf area index (LAI) of winter wheat. The experiment was conducted based on unmanned aerial vehicles (UAV). Shanyang District in the southeast of Jiaozuo City, Henan Province was selected as the experimental area, and LAI data of winter wheat during two growth periods were measured. Firstly, an RGB vegetation index model was constructed, and the optimal VMG model was selected to invert LAI of winter wheat. Then, the sensitivity of PROSAIL parameters was analyzed to obtain the optimal parameter value and invert winter wheat LAI. Finally, the two models were combined using the very fast simulated annealing (VFSA) algorithm to obtain the optimal LAI of winter wheat. The results showed that VFSA can effectively combine PROSAIL model and VMG model to improve the inversion accuracy, and it was better than thta of VMG model and PROSAIL model. The coefficient of determination (R2) was higher than 0.8, and the root mean square error (RMSE) was lower than 0.4m2/m2. To sum up, the ground coverage was increased during the growth of winter wheat, and the method presented had strong radiative transmission mechanism, providing an effective inversion method for LAI inversion.