Data driven modeling under irregular sampling

By calculating the matrix exponential with inverse Laplace transform, the irregular sampling time is transformed to the time-varying parameters. Based on statistics relation between the autocorrelation function and the covariance of Markov random processing, this paper develops a model with adaptive parameters on line. The developed model is used to track target for the video tracking by the common Kalman filter. Simulations and experiments show the good fast-tracking performance get by the model developed here, and it can adaptively adjust the model parameters while tracking and obtain good estimation performance even at very high irregular rate of measurement sampling time.