论文部分内容阅读
针对红外焦平面阵列(IRFPA)探测器盲元和非均匀性导致系统性能降低的问题,首先建立红外焦平面阵列的多元正态分布时序噪声模型,将盲元看作是不符合模型统计分布特性的异常像素点,游离于多元正态分布超椭球之外。其次对序列图像进行主成分分解,将统计距离与等分线空间角作为异常像素检测的统计判据。最后,利用红外热像仪采集了黑体的序列图像数据,用于盲元检测算法的性能验证,实验结果证明该算法的有效性。
In order to reduce the system performance due to blind and inhomogeneity of IRFPA detector, a multivariate normal distribution time series noise model of infrared focal plane array is established, and blind source is considered as not conforming to the statistical distribution of the model Of the abnormal pixels, free from the hyperellipsoid multivariate normal distribution. Secondly, the principal component analysis of the sequence image is adopted, and the statistical distance and the bisector line space angle are taken as the statistical criterion for detecting abnormal pixels. Finally, the sequence data of the blackbody was collected by the infrared thermal imager, which was used to verify the performance of the blind-element detection algorithm. The experimental results show the effectiveness of the algorithm.