We consider the problem of joint carrier frequency offset and channel estimation between transmitter and receiver in a frequency-selective channel MIMO-OFDM system. Recently two high performance estimators based on the expectation-maximization (EM) algorithm have been proposed. The main drawback of the maximum likelihood base algorithms, like EM algorithm, is the high computational complexity. In this paper, we propose an extended Kalman filter based estimator, which has higher performance than that of EM algorithm, while its computational complexity is lower. In addition, the Particle Swarm Optimization (PSO) algorithm is used for joint ML estimation of carrier frequency offset and channel parameters for the general model. The proposed method has a lower computational complexity than that of traditional search methods. Simulation results in comparison with Cramer-Rao bound show that the proposed algorithms outperform the EM in all ranges of signal-to-noise ratio for both channel and frequency offset estimations. Also, among them, the PSO algorithm is superior.