The invention relates to a KFSTRCF-based target tracking architecture, which comprises a discrete time Kalman estimator DKE and an STRCF, and the DKE comprises a discrete time system measurement subsystem and a discrete time system copy subsystem; the target tracking result output of the STRCF serves as measured value input of an observation model in discrete time system measurement, a DKE model is updated through a discrete time system copy subsystem, and a state estimation observation updating equation is obtained. According to the method, the Kalman filter and the STRCF are combined to realize visual tracking, so that the problem of instability caused by large-scale application change is solved; and moreover, a step length control method is also introduced to limit the maximum amplitudeof the output state of the proposed framework, so that the problem of target loss caused by sudden acceleration and steering is solved, the KFSTRCF-based target tracking architecture is superior to STRCF in most cases, and particularly, in sports events, the method shows better performance and stronger robustness than competitors.