基于区域校正的大面阵红外探测器非均匀性校正方法

张明杰; 李岩; 马文坡; 刘紫莹

基于区域校正的大面阵红外探测器非均匀性校正方法

北京空间机电研究所，北京 100094

详细信息

作者简介:
张明杰（1995-），男，硕士研究生，主要从事红外图像处理技术。E-mail：2361636567@qq.com

通讯作者:
马文坡（1967-），男，博士，博士生导师，研究员，主要从事航天光学遥感技术

中图分类号: TN215
计量
- 文章访问数: 391
- HTML全文浏览量: 179
- PDF下载量: 113
- 被引次数: 0
出版历程
- 收稿日期: 2020-10-14
- 修回日期: 2021-04-06
- 刊出日期: 2021-04-20

Non-uniformity Correction for Large Format Array Infrared Detectors Based on Regional Correction

Beijing Institute of Space Mechanics & Electricity, Beijing 100094, China

摘要

摘要: 通过分析某大面阵红外探测器的响应特性，发现了由于相机自身特性引发的不同区域的响应非线性问题。传统的两点校正法或非线性曲线拟合办法对该大面阵探测器校正后，校正残差和目视效果都比较差。本文根据探测器的非线性响应特性，将整个面阵分成了8个区域分别进行非线性拟合校正，然后校正各个区域的偏置系数，最后利用改进的BP神经网络非均匀性校正算法处理区域划分引发的不均匀问题。校正后的各个黑体温度图像的残余非均匀性在千分之一量级，空间噪声也已经十分接近或者小于时间噪声；局部残余非均匀性达到0.002以下，空间噪声明显小于时间噪声。
- 大面阵探测器 /
- 响应非线性 /
- 分区域校正 /
- 神经网络算法
Abstract: By analyzing the response characteristics of large-format infrared detectors, we found that the response nonlinearity in different areas is caused by the characteristics of the camera itself. After the traditional two-point correction method or nonlinear curve fitting method corrects the large-format array detector, the correction residuals and visual effects are relatively poor. In this study, on the basis of eliminating blind elements, according to the nonlinear response characteristics of the detector, the entire array is divided into eight regions for nonlinear fitting correction, the bias coefficient of each region was corrected, and the non-uniformity correction algorithm of the BP neural network was used to deal with the non-uniformity problem caused by region division. After correction, the residual non-uniformity of each black body temperature point image was on the order of one-thousandth, the spatial noise was already very close to or smaller than the temporal noise, and the local residual non-uniformity reached below 0.002, which was significantly smaller than the temporal noise.
- large format array detector /
- response nonlinearity /
- regional correction /
- neural network algorithm

HTML全文

图 1 校正前黑体图像和直方图

Figure 1. Blackbody image and histogram before correction

下载: 全尺寸图片幻灯片

图 2 两点校正后黑体图像和直方图

Figure 2. Blackbody image and histogram after two-point correction

下载: 全尺寸图片幻灯片

图 3 二次曲线拟合算法校正后的黑体图像和直方图

Figure 3. Blackbody image and histogram after quadratic curve fitting correction

下载: 全尺寸图片幻灯片

图 4 分区域校正算法流程图

Figure 4. Flow chart of regional correction algorithm

下载: 全尺寸图片幻灯片

图 5 不同通道两列像元二次拟合响应曲线分析

Figure 5. Quadratic fitting response curves analysis of two columns of pixels in different channels

下载: 全尺寸图片幻灯片

图 6 不同通道两列像元二次项拟合系数差异

Figure 6. Difference of quadratic fitting coefficients of two columns of pixels in different channels

下载: 全尺寸图片幻灯片

图 7 某一行像元的二次项拟合系数

Figure 7. Quadratic fitting coefficient of a row of pixels

下载: 全尺寸图片幻灯片

图 8 盲元检测算法程序流程图

Figure 8. Flow chart of blind element detection algorithm program

下载: 全尺寸图片幻灯片

图 9 死像元剔除前后响应特性比较

Figure 9. Comparison of response characteristics before and after dead pixels are removed

下载: 全尺寸图片幻灯片

图 10 过热像元剔除前后的稳定性比较

Figure 10. Comparison of stability before and after overheated pixels are removed

下载: 全尺寸图片幻灯片

图 11 分区域校正后黑体图像和直方图

Figure 11. Black body image and histogram after regional correction

下载: 全尺寸图片幻灯片

图 12 含有特定周期性噪声的理想图像

Figure 12. Ideal image with particular periodic noise

下载: 全尺寸图片幻灯片

图 13 改进神经网络算法的期望图像

Figure 13. Expected image of improved neural network algorithm

下载: 全尺寸图片幻灯片

图 14 校正后不同通道两列像元二次拟合响应曲线分析

Figure 14. Quadratic fitting response curve analysis of two columns of pixels in different channels after correction

下载: 全尺寸图片幻灯片

图 15 本文方法校正后黑体图像和直方图

Figure 15. Black body image and histogram after correction by this paper

下载: 全尺寸图片幻灯片

表 1 不同算法校正前后图像质量比较

Table 1. Comparison of image quality before and after correction with different algorithms

Temperature/K	Temporal noise(DN value)	Two-point correction algorithm		Quadratic curve fitting correction algorithm		Correction algorithm in this paper
Temperature/K	Temporal noise(DN value)	Spatial noise(DN value）	Residual non-uniformity	Spatial noise(DN value)	Residual non-uniformity	Spatial noise(DN value)	Residual non-uniformity
290.7	12.533	38.64	1.74%	27.13	1.22%	17.07	0.80%
304.1	12.660	21.74	0.80%	20.45	0.75%	11.83	0.45%
313.3	12.773	-	-	-	-	-	-
320.4	12.907	19.86	0.51%	16.91	0.43%	9.48	0.25%
335.5	13.266	72.15	1.25%	32.22	0.56%	11.46	0.20%
346.1	13.623	120.69	1.54%	-	-	-	-
354.5	13.943	168.18	1.68%	82.13	0.82%	20.14	0.20%
359.2	14.038	92.15	0.78%	159.9	1.36%	24.40	0.23%
363.5	13.600	-	-	-	-	-	-

下载: 导出CSV

表 2 不同算法校正前后局部图像质量比较（320.4 K）

Table 2. Comparison of local image quality before and after correction with different algorithms(320.4 K)

Size of region	Temporal noise(DN value)	Two-point correction algorithm		Quadratic curve fitting correction algorithm		Correction algorithm in this paper
Size of region	Temporal noise(DN value)	Spatial noise(DN value)	Residual non-uniformity	Spatial noise(DN value)	Residual non-uniformity	Spatial noise(DN value)	Residual non-uniformity
11×11	12.92	5.52	0.14%	8.76	0.22%	4.83	0.13%
51×51	12.13	11.26	0.29%	9.76	0.25%	6.29	0.16%
101×101	12.19	12.73	0.32%	10.02	0.25%	7.09	0.18%

下载: 导出CSV

参考文献(9)

[1]	Rogalski A. Next decade in infrared detectors[C]//Electro-Optical and Infrared Systems: Technology and Applications XIV, 2017: 104330L (https://doi.org/10.1117/12.2300779).
[2]	任建乐, 陈钱, 钱惟贤, 等. 基于多帧配准的红外焦平面阵列非均匀性自适应校正[J]. 红外与毫米波学报, 2014, 33(2): 122-128. https://www.cnki.com.cn/Article/CJFDTOTAL-HWYH201402002.htm REN Jianle, CHEN Qian, QIAN Weixian, et al. Multiframe registration based adaptive nonuniformity correction algorithm for infrared focal plane arrays[J]. Journal of Infrard and Millimeter Waves, 2014, 33(2): 122-128. https://www.cnki.com.cn/Article/CJFDTOTAL-HWYH201402002.htm
[3]	钱润达, 赵东, 周慧鑫, 等. 基于加权引导滤波与时域高通滤波的非均匀性校正算法[J]. 红外与激光工程, 2018, 42(12) : 1-6. https://www.cnki.com.cn/Article/CJFDTOTAL-HWYJ201812022.htm QIAN Runda, ZHAO Dong, ZHOU Huixin, et al. Non-uniformity correction algorithm based on weighted guided filter and temporal high-pass filter[J]. Infrared and Laser Engineering, 2018, 42(12) : 1-6. https://www.cnki.com.cn/Article/CJFDTOTAL-HWYJ201812022.htm
[4]	陈钱, 隋修宝. 红外图像处理理论与技术[M]. 北京: 电子工业出版社, 2018. CHEN Qian, SUI Xiubao. Infrared Image Processing Theory and Technology[M]. Beijing: Publishing House of Electronics Industry, 2018.
[5]	吕游, 何昕, 魏仲慧. 红外焦平面阵列非均匀性校正算法研究[J]. 计算机技术与发展, 2015, 25(2): 1-5. https://www.cnki.com.cn/Article/CJFDTOTAL-WJFZ201502001.htm LV You, HE Xin, WEI Zhonghui. Research on non-uniformity correction algorithms for IRFPA[J]. Computer Tecnology and Development, 2015, 25(2): 1-5. https://www.cnki.com.cn/Article/CJFDTOTAL-WJFZ201502001.htm
[6]	李成立, 吕俊伟, 王佩飞, 等. 红外探测器盲元检测及评价[J]. 激光与红外, 2018, 48(2): 209-214. doi: 10.3969/j.issn.1001-5078.2018.02.014 LI Chengli, LV Junwei, WANG Peifei, et al. Blind pixel detection and evaluation for infrared detector[J]. Laser and Infrared, 2018, 48(2): 209-214. doi: 10.3969/j.issn.1001-5078.2018.02.014
[7]	国家质量监督检验检疫总局, 国家标准化管理委员会. GB/T17444 -2013红外焦平面阵列参数测试方法[S]. 中国标准出版社, 2014. General Administration of Quality Supervision, Inspection and Quarantine of China, Standardization Administration. GB/T17444-2013 Measuring methods for parameters of infrared focalplane arrays[S]. China Standards Press, 2014.
[8]	Dean A Scribner, Kenneth A Sarkady, Melvin R Kruer, et al. Adaptive nonuniformity correction for IR focal-plane arrays using neural networks[C]// Proc. of Infrared Sensors: Detectors, Electronics, and Signal Processing, 1991, 1541: 100-109.
[9]	张丽莎, 刘兆军. 基于90°旋转定标和场景校正相结合的非均匀性校正技术[J]. 航天返回与遥感, 2017, 38(1): 78-87. https://www.cnki.com.cn/Article/CJFDTOTAL-HFYG201701011.htm ZHANG Lisha, LIU Zhaojun. High performance NUC by side-slither combined with scened-based correction[J]. Spacecraft Covery And Remote Sensing, 2017, 38(1): 78-87. https://www.cnki.com.cn/Article/CJFDTOTAL-HFYG201701011.htm