制冷型红外焦平面探测器的响应非均匀性对其定量化应用产生很多不利影响。本文在对红外焦平面探测器像元响应特性深入研究的基础上，提出了一种的改进型自适应分段的红外焦平面非均匀校正算法。该方法主要在前人研究的基础上给出校正效果与分段校正阈值选择之间的定量化关系，从而降低了分段校正时阈值选择的盲目性。另外，该方法采用归一化直线作为校正的目标曲线，而非传统方法中采用分段目标曲线。文中首先分析了线性校正误差的来源，然后给出校正效果与分段阈值之间的定量化关系，最后给出校正方法的原理和步骤。在校正效果评价方面，本文使用局部非均匀性(LNU)来评价校正后的效果，该方法相对于传统的评价方法，更能体现校正后残差的局部特性。实验表明，相对于传统的多点校正法，该方法使用较少的参数量获得更好的校正效果，并且易于工程实现。

The response non-uniformity of the cooled infrared focal plane detector has a lot of adverse effects on its quantitative application. Based on the study of the pixel response characteristics of the cooled infrared focal plane detector, this paper presents an improved adaptive segmentation non-uniformity correction method for IRFPA. On the basis of previous studies, this method gives the quantitative relation between the correction effect and the threshold selection of segmentation correction, which can reduce the blindness of the threshold selection in segmentation correction. In addition, the method uses the normalized straight line as the target curve, which is different from the traditional methods. This paper analyzes the source of the linear correction error firstly, and then presents the quantitative relation between the correction effect and the segmentation threshold. Finally, the principle and the steps of the correction method are given. In the correction effect evaluation, this paper uses the local non-uniformity (LNU) to evaluate the effect of the correction, and the method is better than the traditional evaluation method, which can reflect the local characteristics of the residual after correction. Experiments show that this method uses less parameters to obtain better correction effect, compared with the traditional multi point correction method, and easy to be realized.