【作者】 李志华；

【导师】 王兰勋；

【作者基本信息】 河北大学 ， 通信与信息系统， 2004， 硕士

【摘要】 红外焦平面应用于红外成像系统具有结构简单、灵敏度高、功耗小等优点，所以得到高度重视和广泛应用。然而，红外焦平面阵列固有的非均匀性，使红外成像系统的温度分辨率下降，图像质量受到严重影响。因此，必须对红外焦平面进行非均匀性校正。 本论文首先对红外焦平面非均匀性的产生原因进行了探讨，而后选取了便于实际操作的非均匀性定义及大小度量方式，并介绍了几种非均匀性校正算法及其优、缺点。在此基础上，指出自适应非均匀性校正方法是非均匀性校正发展的必然趋势。进而详细介绍了自适应非均匀性校正方法中最具发展前景的人工神经网络校正算法，并对各种人工神经网络算法进行了研究和比较，选取BP神经网络应用于红外焦平面的非均匀性校正。参阅国外文献，给出了的非均匀性BP校正算法的实际应用模型。通过对经典BP校正算法模型的分析研究，针对经典BP校正算法参数难以选取的缺点，提出了一种改进型的BP校正算法——归一化BP校正算法的实现方式，在参数选取的合理性和可行性方面得到了明显的改善。分析归一化BP校正算法的参数选取与算法敛散性之间的关系，对实际应用中参数的选取具有指导作用。对改进的归一化BP校正算法利用Matlab进行了软件仿真，在取得良好的非均匀性校正效果的同时，验证了对归一化BP校正算法理论分析的正确性。最后对算法的硬件可实现性进行了论证。更多还原

【Abstract】 Infrared focal plane array(IRFPA) can be superior for infrared imaging systems not only because of its simplicity and compactness in structure ,but also because of its high sensitivity and low power depletion. And infrared focal plane array is applied to imaging system widely and attracted lots of attention. Unfortunately, its inherent nonuniformity can degrade image quality and system performance. Hence, nonunifbrmity correction(NUC) becomes necessary.The paper firstly describes the reasons of the nonuniformity in the IRFPA. Then several methods of nonuniformity correction(NUC) for IRFPA are analyzed and compared, which leads to that the adaptive correction scheme is the best for the IR imaging system. Following that, artificial neural networks correction technique-back propagation(BP) algorithm is discussed in detail. On the base of above analysis, the unitary BP algorithm is introduced, which has more adaptability than classical BP algorithm. Experiments were performed. The unitary BP algorithm’s effectiveness and superiority were verified by simulation. At the end of this paper, the possibility of hardware-realization of the algorithm is validated.更多还原