Estimation of treatment effects in observational studies by recovering the assignment probabilities and the population model