Dissipative particle dynamics (DPD) is a meso-scale, Lagrangian particle method, and has been successfully applied to different areas including micro- and nano-fluidics, bio- and chemical technologies. The treatment of solid matrix and the implementation of solid boundary conditions have been an important task for the development and application of the DPD method. This paper presents a new method of treating complex solid boundary. Solid grains in complex flow geometry can be represented by freezing randomly distributed DPD particles which have reached an equilibrium state. To increase computational efficiency, only the boundary DPD particles within one cut-off distance from the flow region are frozen. A thin layer in the flow region next to the solid boundary is used to bounce mobile DPD particles in this layer back to the flow region. The DPD method and this new boundary treatment algorithm are used to model the Poiseuille flow and a flow problem in a complex porous media. It is demonstrated that this new boundary treatment algorithm can effectively model complex solid matrix and correctly implement non-slip boundary condition.