In this paper, we present a modified discrete Fourier transform (DFT)-based channel estimator for orthogonal frequency-division multiplexing (OFDM) systems with null carriers. Due to the channel leakage, the conventional DFT-based estimator has the problem of error floor. In order to suppress the leakage, we adopt the optimal leakage suppression method. Moreover, a novel denoising method based on dual-tree complex wavelet is proposed to further improve the performance. Simulation results show that the proposed channel estimator can effectively overcome the performance degradation caused by the leakage and additive-white gaussian noise (AWGN).