二维光谱图像的神经网络校正方法

doi:10.3778/j.issn.1673-9418.2201072

摘要/Abstract

摘要： 二维光谱图像普遍存在畸变现象，这种畸变会影响抽谱操作，从而降低一维光谱数据的质量。针对上述问题，提出了一种有效的基于神经网络的二维光谱图像的校正方法。该方法首先从二维光谱图像中提取表示畸变特征的数据，即抽取平场光谱中每条光纤的中心线以及拟合定标灯谱中每个特定波长处的特征谱线，再根据这两组线构造数据集；然后设计神经网络模型，训练模型拟合图像校正前后的像素坐标值之间的关系，通过模型计算出校正图像的所有像素坐标值；最后根据原始畸变图像的通量值一一对应填充校正图像的通量值。分别用平场光谱、定标灯谱、目标光谱数据进行校正实验。与此同时，对比了目标光谱图像校正前后的抽谱结果。实验证明该方法能够有效校正弯曲的二维光纤光谱图像，并在一定程度上提升一维光谱数据的质量。

关键词: 二维光谱, 神经网络, 图像处理, 图像校正

Abstract: Two-dimensional spectral images are generally distorted. Spectrum extraction operation is affected by such distortion, which reduces the quality of one-dimensional spectral data. Aiming at above problem, an effective correction method for the distorted two-dimensional spectral images based on neural network is proposed. Firstly, by extracting the center line of each fiber from the flat-field spectrum and fitting the equal-wavelength line at each specific wavelength from the calibration lamp spectrum, data that represent distortion characteristics from two-dimensional spectral images can be obtained. The training samples are thus constructed according to these two sets of feature lines. Secondly, a neural network model is then designed and trained to fit the relation between the pixel coordinates of the image before and after correction. Therefore, all pixel coordinate values of the corrected image can be calculated by the model. Finally, the flux values of the corrected image are filled one-to-one in accord with the flux value of the original distorted image. The correction experiments are carried out with the flat-field spectrum, calibration lamp spectrum, and object spectrum respectively. The spectral extraction results of the object spectrum before and after correction are compared. Experimental results prove that the method can correct the distorted two-dimensional spectral image effectively and improve quality of one-dimensional spectral data to an extent.

Key words: two-dimensional spectra, neural network, image processing, image correction

尹乾, 王燕, 郭平, 郑新. 二维光谱图像的神经网络校正方法[J]. 计算机科学与探索, 2023, 17(7): 1622-1633.

YIN Qian, WANG Yan, GUO Ping, ZHENG Xin. Distortion Correction of Two-Dimensional Spectral Image Based on Neural Network[J]. Journal of Frontiers of Computer Science and Technology, 2023, 17(7): 1622-1633.

参考文献

[1] 邓小超, 王坚, 董健, 等. LAMOST CCD相机总控的设计与实现[J]. 天文研究与技术, 2011, 8(3): 279-284.
DENG X C, WANG J, DONG J, et al. Design and imple-mentation of the general control of the CCD cameras on the LAMOST[J]. Astronomical Research & Technology, 2011, 8(3): 279-284.
[2] 秦瀚钦. LAMOST二维数据处理方法研究[D]. 北京: 中国科学技术大学, 2010.
QIN H Q. Research on two-dimensional data processing method of LAMOST[D]. Beijing: University of Science and Technology of China, 2010.
[3] 张博, 叶中付, 徐旭. 一种基于二维算法的新颖的多目标光纤光谱数据处理流程[J]. 天文学报, 2016, 57(1): 91-101.
ZHANG B, YE Z F, XU X. A novel spectral data proces-sing procedure on multi-object fiber spectral data based on 2-D algorithms[J]. Acta Astronomica Sinica, 2016, 57(1): 91-101.
[4] 汪梦欣, 陈笑然, 罗阿理, 等. 前照条件下LAMOST焦面板光纤单元定位精度检测系统设计[J]. 天文研究与技术, 2014, 11(2): 157-164.
WANG M X, CHEN X R, LUO A L, et al. Design of a system for measuring positioning accuracies of optical-fiber units on the LAMOST focal plane using front light sources[J]. Astronomical Research and Technology, 2014, 11(2): 157-164.
[5] RAHMAN T, KROUGLICOF N. An efficient camera cali-bration technique offering robustness and accuracy over a wide range of lens distortion[J]. IEEE Transactions on Image Processing, 2012, 21(2): 626-637.
[6] BUKHARI F, DAILEY M N. Automatic radial distortion esti-mation from a single image[J]. Journal of Mathematical Imaging & Vision, 2013, 45(1): 31-45.
[7] MIROSLAV G, JESSICA F. Estimation of lens distortion cor-rection from single images[J]. Proceedings of SPIE the Inter-national Society for Optical Engineering, 2014, 8(2): 1267-1277.
[8] 赵嘉毅, 祝恩, 印万鹏. 接触网测量中线阵摄像机标定的SVR方法[J]. 计算机科学与探索, 2018, 12(11): 1788-1795.
ZHAO J Y, ZHU E, YIN W P. Linear array camera calibra-tion method based on support vector regression for contact wire measurements[J]. Journal of Frontiers of Computer Science and Technology, 2018, 12(11): 1788-1795.
[9] 朱海静, 邱波, 陈建军, 等. 基于法向映射法的二维光纤光谱弯曲校正[J]. 光谱学与光谱分析, 2019, 39(2): 622-627.
ZHU H J, QIU B, CHEN J J, et al. Bending correction of 2D optical fiber spectra based on normal mapping method[J]. Spectroscopy and Spectroscopy Analysis, 2019, 39(2): 622-627.
[10] 郑子鹏, 邱波, 魏诗雅, 等. 曲线距离法的二维光纤光谱弯曲校正[J]. 光谱学与光谱分析, 2019, 39(10): 3051-3055.
ZHENG Z P, QIU B, WEI S Y, et al. Two-dimensional fiber spectral bending correction based on curve distance method[J]. Spectroscopy and Spectral Analysis, 2019, 39 (10): 3051-3055.
[11] 杨滨, 陈先意. 多数据集深度学习模型的修图处理识别[J]. 计算机科学与探索, 2020, 14(2): 252-259.
YANG B, CHEN X Y. Image modification recognition based on multi-dataset deep learning model[J]. Journal of Frontiers of Computer Science and Technology, 2020, 14(2): 252-259.
[12] 江祥奎, 范永青, 王婉. 基于粒子群遗传算法的BP神经网络摄像机标定[J]. 计算机科学与探索, 2014, 8(10): 1254-1262.
JIANG X K, FAN Y Q, WANG W. BP neural network camera calibration based on particle swarm optimization genetic algorithm[J]. Journal of Frontiers of Computer Science and Technology, 2014, 8(10): 1254-1262.
[13] 王珂娜, 邹北骥, 黄文梅. 一种基于神经网络的畸变图像校正方法[J]. 中国图象图形学报, 2005, 10(5): 603-607.
WANG K N, ZOU B J, HUANG W M. A distorted image correction method based on neural network[J]. Chinese Journal of Image and Graphics, 2005, 10(5): 603-607.
[14] 王牧云, 严壮志, 葛俊杰. 医用内窥镜图像校正的 BP 神经网络方法[J]. 应用科学学报, 2009, 27(5): 480-484.
WANG M Y, YAN Z Z, GE J J. Distortion correction of medical endoscopic images using BP neural network[J]. Journal of Application Science, 2009, 27 (5): 480-484.
[15] 张峰峰, 杜志江, 孙立宁. 基于BP神经网络的X光图像畸变校正技术的研究[J]. 计算机应用研究, 2008, 25(10): 3050-3052.
ZHANG F F, DU Z J, SUN L N. Research of correcting distorted X-ray image based on BP neural network[J]. App-lication Research of Computers, 2008, 25(10): 3050-3052.
[16] LI M Z, HAERKEN H, GUO P, et al. Two-dimensional spec-tral image calibration based on feed-forward neural network[C]//Proceedings of the 2017 International Joint Conference on Neural Networks, Anchorage, May 14-19, 2017. Piscata-way: IEEE, 2017: 4194-4201.
[17] HANEISHI H, YAGIHASHI Y, MIYAKE Y. A new method for distortion correction of electronic endoscope images[J]. IEEE Transactions on Medical imaging, 1995, 14(10): 548-555.
[18] 梁亮, 张定华, 赵歆波, 等. 一种基于网格图像的几何畸变修正方法[J]. 计算机工程与应用, 2004, 40(2): 4-6.
LIANG L, ZHANG D H, ZHAO X B, et al. An image geometry calibration method based on grid-image[J]. Com-puter Engineering and Applications, 2004, 40(2): 4-6.
[19] 朱铮涛, 黎绍发, 陈华平. 基于平面网格模型的摄像机镜头畸变校正技术[J]. 计算机工程与应用, 2004, 40(23): 32-35.
ZHU Z T, LI S F, CHEN H P. Technology of proofreading the camera lens distortion based on the plane gridding model[J]. Computer Engineering and Applications, 2004, 40(23): 32-35.
[20] MARKOVSKY I, VAN H S. Overview of total least-squares methods[J]. Signal Processing, 2007, 87(10): 2283-2302.
[21] MARKOVSKY I, SIMA D M, HUFFEL S V. Total least squares methods[J]. Wiley Interdisciplinary Reviews Com-putational Statistics, 2010, 2(2): 212-217.
[22] ZEILER M D. ADADELTA: an adaptive learning rate me-thod[J]. arXiv:1212.5701, 2012.
[23] KINGMA D P, BA J. Adam: a method for stochastic opti-mization[J]. arXiv:1412.6980, 2014.
[24] 张博. LAMOST光谱数据的二维处理关键技术研究[D]. 中国科学技术大学, 2016.
ZHANG B. Research on the key technology of 2D pro-cessing of LAMOST spectral data[D]. University of Science and Technology of China, 2016.
[25] QIN H Q, YE Z F, LUO A. Flux extraction based on radial basic function for LAMOST[J]. Publications of the Astrono-mical Society of the Pacific, 2009, 121(878): 408-413.