A discrete-time filtering problem for hybrid systems is studied in which the state process is observed in white noise and the random jump process (called the mode or the regime) is observed by a point process. The optimal state and mode estimator, when only the mode measurements are used, is studied. Some properties of this filter are then established. Apart from their theoretical interest, these results have an interesting practical implication for the design of image-based tracking systems, since they lead to an easily implementable suboptimal algorithm. Two illustrative examples are given to show the performance of the proposed approach.