The existing opportunistic network routing algorithms assume that nodes do not exhibit selfish behavior. In a well-cooperating network, these algorithms can effectively deliver messages. However, when nodes behave selfishly, the efficiency of these algorithms decreases. To address this issue, this paper proposes a Filtering Algorithm of Selfish Node(FASN).This algorithm filters selfish nodes by reducing the transmission probability of selfish nodes. Additionally, to alleviate network congestion, a new ACK confirmation mechanism is proposed. This mechanism not only removes redundant messages that have already been delivered but also deletes messages with a survival rate of 0. Finally, node differentiation punishment is achieved based on the number of times a node is added to a blacklist. Experimental results demonstrate that the FASN algorithm is suitable for opportunistic networks with high security requirements. Moreover, it effectively filters out selfish nodes and outperforms the CDEC algorithm and several classical routing algorithms in terms of delivery rate, overhead, and average delay.