Xu Jiangtao, Nie Huafeng, Nie Kaiming, Jin Weimin
J Opt Soc Am A Opt Image Sci Vis. 2017 Sep 1;34(9):1500-1510. doi: 10.1364/JOSAA.34.001500.
In this paper, an improved moment matching method based on a spatial correlation filter (SCF) and bilateral filter (BF) is proposed to correct the fixed-pattern noise (FPN) of a time-delay-integration CMOS image sensor (TDI-CIS). First, the values of row FPN (RFPN) and column FPN (CFPN) are estimated and added to the original image through SCF and BF, respectively. Then the filtered image will be processed by an improved moment matching method with a moving window. Experimental results based on a 128-stage TDI-CIS show that, after correcting the FPN in the image captured under uniform illumination, the standard deviation of row mean vector (SDRMV) decreases from 5.6761 LSB to 0.1948 LSB, while the standard deviation of the column mean vector (SDCMV) decreases from 15.2005 LSB to 13.1949LSB. In addition, for different images captured by different TDI-CISs, the average decrease of SDRMV and SDCMV is 5.4922/2.0357 LSB, respectively. Comparative experimental results indicate that the proposed method can effectively correct the FPNs of different TDI-CISs while maintaining image details without any auxiliary equipment.
本文提出了一种基于空间相关滤波器（SCF）和双边滤波器（BF）的改进矩匹配方法，用于校正延时积分互补金属氧化物半导体图像传感器（TDI - CIS）的固定模式噪声（FPN）。首先，分别通过SCF和BF估计行FPN（RFPN）和列FPN（CFPN）的值，并将其添加到原始图像中。然后，对滤波后的图像采用带移动窗口的改进矩匹配方法进行处理。基于128级TDI - CIS的实验结果表明，在对均匀光照下采集的图像校正FPN后，行均值向量的标准差（SDRMV）从5.6761 LSB降至0.1948 LSB，而列均值向量的标准差（SDCMV）从15.2005 LSB降至13.1949 LSB。此外，对于不同TDI - CIS采集的不同图像，SDRMV和SDCMV的平均降幅分别为5.4922/2.0357 LSB。对比实验结果表明，该方法无需任何辅助设备，即可有效校正不同TDI - CIS的FPN，同时保留图像细节。
