红外焦平面阵列无效像元检测技术研究

Study of IRFPA Non-effective Pixel Discrimination

提出了一种基于噪声等效温差(NETD)和灰度最小方差的红外焦平面阵列无效像元检测方法。无效像元包括过热像元和死像元。首先利用三维噪声模型建立图像数据集,然后分别判断过热像元和死像元。通过像元的噪声电压值与10倍NETD的比较来实现过热像元的判别;对于死像元的判别处理是,设计了9个不同的窗口,分别计算每个窗口的方差并选取方差最小的窗口,最后通过比较窗口中心像元灰度与均值灰度的差值是否落在3倍方差内作为评判死像元的标准。最后通过实验证明,该方法具有误判率较小,定位准确率高等特点。

A method of IRFPA non-effective pixel discrimination based on NETD and 8ray minimum variance is put forward. First, data sets are established by using 3-D noise model. Second,over hot pixels and dead pixels are distinguished individually. By comparing pixel noise voltage and 10 times of NETD, over hot pixels are got. Nine windows are created, and then mean value and standard error of every window is calculated and the window with the minimized standard error is chosen. At last we distinguish the dead pixel by using ＂3-D＂principle from the minimized window. The result shows that this method is practical and acceptable.