Abstract: A traffic violation reporting system was designed based on swarm intelligence perception, and the existing data selection mechanisms were analyzed. In view of the shortcomings of these mechanisms in evaluating the perceived data quality and the credibility of users, a new data selection method based on reputation model was proposed.The user's reputation value was dynamically calculated according to the quality of the historical perceived data. This method was applied to the design of the violation reporting system, which could effectively limit the malicious reporting behavior and improve the quality of the reporting data and the system operation efficiency. According to the actual test, the system operated stably and reliably, which was a good supplement to the existing traffic monitoring system.