On the enhancement of high-gain observers for state estimation of nonlinear systems