An effective total variation-based nonuniformity correction (NUC) is presented to remove the nonuniformity in infrared focal plane array in this paper. In this method, we treat the NUC problem as an ill-posed inverse problem and set L1-norm horizontal total variation regularization (L1-HTV) to solve. By a more realistic geometrical consideration on the nonuniform noise, the L1-HTV regularization only constrains the gradients along the x-axis in space domain and time domain while protects the gradients along the y-axis. As a result, nonuniformity can perfectly removed effectively, while the details and edges are well preserved. In addition, we use the Split Bregman iteration to minimize the regularization term in the model, which highly reduces the computational complexity so that it can be used in real-time video sequence. We test the experiments on real scene under different circumstances and indicate that our algorithm is effective and efficient.