This paper proposes a reflectivity constraint for velocity estimation to optimally solve the inverse problem for active seismic imaging. This constraint is based on the velocity model derived from the definition of reflectivity and acoustic impedance. The constraint does not require any prior information of the subsurface and large extra computational costs, like the calculation of so-called Hessian matrices. We incorporate this constraint into the joint migration inversion algorithm, which simultaneously estimates both the reflectivity and velocity model of the subsurface in an iterative process. Using so-called full wavefield modelling, the misfit between forward modelled and measured data is minimized. Numerical and field data examples are given to demonstrate the validity of our proposed algorithm in case accurate initial models and the low-frequency components of observed seismic data are absent.