The classification performance of brain-computer interface (BCI) depends on the preprocessing method of EEG signals to some extent. In this study, we propose a preprocessing method for spatiotemporal filtering to solve the effect of latency delay in the human visual system on target recognition performance of code modulated visual evoked potential (c-VEP) BCI. A spatiotemporal filter was created based on the least mean square error (LMSE) between an average signal and a single test signal, and a sparsity constraint was applied to the weight coefficient of the filter by the least absolute shrinkage and selection operator (LASSO), and Template matching method was used to identify the target. The algorithm was applied to the c-VEP BCI of 16 targets modulated by the 63-bit M-sequence and its cyclic shift sequence, and compared with the conventional canonical correlation analysis (CCA) spatial filtering algorithm and spatiotemporal inverse filtering algorithm. The results of twelve subjects show that our proposed algorithm outperforms other two algorithms in classification accuracy.