The density field reconstruction technique, which partially reverses the non-linear degradation of the Baryon acoustic oscillation (BAO) feature in the galaxy redshift surveys, has been successful in substantially improving the cosmology constraints from recent surveys such as Baryon Oscillation Spectroscopic Survey (BOSS). We estimate the efficiency of the method as a function of various reconstruction details. To directly quantify the BAO information in non-linear density fields before and after reconstruction, we calculate the cross-correlations (i.e. propagators) of the pre(post)-reconstructed density field with the initial linear field using a mock sample that mimics the clustering of the BOSS galaxies. The results directly provide the BAO damping as a function of wavenumber that can be implemented into the Fisher matrix analysis. We focus on investigating the dependence of the propagator on a choice of smoothing filters and on two major different conventions of the redshift-space density field reconstruction that have been used in literature. By estimating the BAO signal to noise for each case, we predict constraints on the angular diameter distance and Hubble parameter using the Fisher matrix analysis. We thus determine an optimal Gaussian smoothing filter scale for the signal-to-noise level of the BOSS CMASS. We also present appropriate BAO fitting models for different reconstruction methods based on the first- and second-order Lagrangian perturbation theory in Fourier space. Using the mock data, we show that the modified BAO fitting model can substantially improve the accuracy of the BAO position in the best fits as well as the goodness of the fits.