In this paper, we investigate the filter design problem for linear continuous-time systems with parameter variations in system matrices. The parameter variations are assumed to belong to a polytope with finite and known vertices. The designed filter parameters and the constructed Lyapunov function are both dependent on the online measured variations. A new sufficient condition for the existence of parameter-dependent filters is established and it can guarantee that the filtering error dynamic system is asymptotically stable and can satisfy the prescribed $\mathcal{H}\u221e$ norm bound. Then, the design of the filter is proposed by solving a set of linear matrix inequalities (LMIs). Simulation studies and comparison examples are provided to illustrate the effectiveness of the proposed method.