针对传统的红外图像非均匀性校正方法精度低，易破坏图像细节和边缘等缺点，本文提出了一种新的基于全变分理论的红外图像非均匀性校正方法。在分析不同正则项对全变分模型去噪性能影响的基础上，针对红外图像条纹非均匀性的几何特征，对原有的全变分模型进行了修正，使新模型既能约束图像水平方向的梯度，又能保护图像垂直方向的梯度。通过Split Bregman 迭代最小化新的全变分模型，显著降低了计算复杂度，使其能广泛应用于实时视频序列。通过不同环境下对真实场景的实验，表明该方法不但能有效地校正红外图像的条纹非均匀性，还能较大程度地保护住图像的细节和边缘信息。

Traditional nonuniformity correction algorithms for infrared focal plane array suffer from drawbacks such as low correction accuracy and damages in image details and edges. Aiming at these problems, a new method based on total variation is proposed for infrared image nonuniformity correction. On the basis of analyzing the influence of different regularization terms on the performance of the total variation model and geometric features of the stripe nonuniformity, we modify the original total variation model so that the new model can either constrain the image gradient of the horizontal direction, or protect the image gradient of the vertical direction. By using the Split Bregman to minimize the total variation model, the computational complexity is significantly reduced 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 can effectively remove the column stripe nonuniformity, while the details and edges are well preserved.