In order to build a proper maneuvering target model for the infrared fisheye warning system, the target moving characteristics on the image surface are analyzed. A modified constant velocity model is presented by virtue of the system′s higher sampling frequency. Comparing with the current statistical model, its tracking accuracy is analyzed theoretically. It is found that, for the infrared fisheye warning system with high sampling frequency, the tracking accuracy of the constant velocity model can reach that of the current statistical model. With Kalman filtering, the analog track is used for the tracking experiment, and it is proved that the result above is correct. Above all, this model, with its simple structure and less computation, is available for the infrared fisheye warning system.