Orthogonal frequency division multiplexing (OFDM) has been considered as a kind of effective means to realize wideband data transmission by applying multiple low rates and parallel data transmission subcarriers. A very important characteristic of wireless channel is multipath propagation, which makes the received signal overlap, and leads to inter symbol interference (ISI). Therefore, OFDM channel estimation is an important part of OFDM system. This paper proposed a channel estimation algorithm based on Kalman filter compression sensing (KF-CS). By the combination of KF and CS algorithm, the KF-CS gets higher estimation precision with fewer pilots compared with classical algorithms. We adopted LS, LMMSE, CS, KF and KF-CS algorithms in pilot insertion method, obtained complete channel spectral response and got effective comparison of mean square error (MSE) of the channel response. Simulation results show that the proposed algorithm of KF-CS has better accuracy than KF algorithm, and can reduce pilots. So the proposed algorithm can get good channel estimation performance with fewer pilots and improved spectrum efficiency.