一种高超星图高精度半盲复原算法
|
摘要
针对高超声速飞行器在平流层内应用天文导航时受气动光学效应及运动模糊影响后难以观星和高精度导航的问题,提出一种基于正则化思想的高超星图半盲复原算法。该算法首先针对高超星图的特点进行去噪与星点初提取等预处理操作,接着从图像中提出可用的模糊核信息,并通过融合达到去噪的目的。然后结合天文图像灰度及梯度的稀疏先验分布特性,提出一种针对高超星图的正则化非盲复原模型,利用分裂布雷格曼迭代法等算法迭代估计清晰图像。将本算法与传统星图复原算法、其他最新正则化复原算法进行星图复原与导航效果比较,结果表明本算法复原效果最佳,且能明显改善星点识别正确率及质心坐标计算精度,可用于大幅提高超声速飞行器在平流层中的天文导航适应性及导航精度。
A semi-blind restoration algorithm of a hypersonic star image based on the regularization method is proposed, which can be used in the celestial navigation of a hypersonic vehicle disturbed by the aero-optical effect and motion blur in order to solve the problem of the star recognition and navigation accuracy. In the algorithm, the pre-processing operations are carried out firstly, such as denoising and star-point initilal extration depend on the characteristics of hypersonic star images. The useful imformation of kernel is extracted from image and fused to denoise the kernel. Based on the sparse prior of both the gray image and its gradient of the star image, a regularized nonblind restoration model for a hypersonic star image is proposed to estimate the clear image with the help of the Bregman iterative algorithm. The algorithm is compared with the traditional star image restoration algorithm and the other state-of-art regularization restoration algorithms in experiments. The results show that the proposed algorithm has the best recovery ability, which can obviously improve the accuracy of star recognition. Thus, it can be used to greatly improve the adaptability and accuracy of the celestial navigation of a supersonic vehicle in the stratosphere.
A semi-blind restoration algorithm of a hypersonic star image based on the regularization method is proposed, which can be used in the celestial navigation of a hypersonic vehicle disturbed by the aero-optical effect and motion blur in order to solve the problem of the star recognition and navigation accuracy. In the algorithm, the pre-processing operations are carried out firstly, such as denoising and star-point initilal extration depend on the characteristics of hypersonic star images. The useful imformation of kernel is extracted from image and fused to denoise the kernel. Based on the sparse prior of both the gray image and its gradient of the star image, a regularized nonblind restoration model for a hypersonic star image is proposed to estimate the clear image with the help of the Bregman iterative algorithm. The algorithm is compared with the traditional star image restoration algorithm and the other state-of-art regularization restoration algorithms in experiments. The results show that the proposed algorithm has the best recovery ability, which can obviously improve the accuracy of star recognition. Thus, it can be used to greatly improve the adaptability and accuracy of the celestial navigation of a supersonic vehicle in the stratosphere.
引文
[1] 殷兴良.气动光学原理[M].北京:中国宇航出版社,2003.
[2] Pond J E,Sutton G W.Aero-optic performance of an aircraft forward-facing optical turret[J].Journal of Aircraft,2015,43(3):600-607.
[3] Gao Q,Yi S H,Jiang Z F,et al.Temporal evolution of optical path difference of a supersonic turbulent boundary layer[J].Chinese Physics B,2013,22(1):014202.
[4] 洪汉玉,张天序.基于各向异性和非线性正则化的湍流退化图像复原[J].宇航学报,2004,25(1):5-12.[Hong Han-yu,Zhang Tian-xu.Restoration of turbulence-degraded images using anisotropic and nonlinear regularization[J].Journal of Astronautics,2004,25(1):5-12.]
[5] 洪汉玉,张天序,余国亮.基于Poisson模型的湍流退化图像多帧迭代复原算法[J].宇航学报,2004,25(6):649-654.[Hong Han-yu,Zhang Tian-xu,Yu Guo-liang.Iterative multi-frame restoration algorithm of turbulence-degraded images based on passion model[J].Journal of Astronautics,2004,25(6):649-654.]
[6] Fergus R,Singh B,Hertzmann A,et al.Removing camera shake from a single photograph[J].Acm Transactions on Graphics,2006,25(3):787-794.
[7] Krishnan D,Fergus R.Fast image deconvolution using hyper-Laplacian priors[C].International Conference on Neural Information Processing Systems,Brussels,Belgium,Sep 11-14,2011.
[8] Ohkoshi K,Goto T,Hirano S,et al.Blind image restoration based on total variation regularization of blurred images[C].The 1st IEEE Global Conference on Consumer Electronics,Tokyo,Japan,Oct 2-5,2012.
[9] Pan J,Sun D,Pfister H,et al.Blind image deblurring using dark channel prior[C].Computer Vision and Pattern Reco-gnition,Las Vegas,USA,Jun 27-30,2016.
[10] Krishnan D,Fergus R.Fast image deconvolution using hyper-Laplacian priors[C].International Conference on Neural Information Processing Systems,Vancouver,Canada,Dec 7-10,2009.
[11] Krishnan D,Tay T,Fergus R.Blind deconvolution using a normalized sparsity measure[C].Computer Vision and Pattern Recognition,Colorado Springs,USA,Jun 20-25,2011.
[12] Pan J,Su Z.Fast L0-regularized Kernel estimation for robust motion deblurring[J].IEEE Signal Processing Letters,2013,20(9):841-844.
[13] Levin A,Weiss Y,Durand F,et al.Understanding and evaluating blind deconvolution algorithms[C].Computer Vision and Pattern Recognition,Miami,USA,Jun 20-25,2009.
[2] Pond J E,Sutton G W.Aero-optic performance of an aircraft forward-facing optical turret[J].Journal of Aircraft,2015,43(3):600-607.
[3] Gao Q,Yi S H,Jiang Z F,et al.Temporal evolution of optical path difference of a supersonic turbulent boundary layer[J].Chinese Physics B,2013,22(1):014202.
[4] 洪汉玉,张天序.基于各向异性和非线性正则化的湍流退化图像复原[J].宇航学报,2004,25(1):5-12.[Hong Han-yu,Zhang Tian-xu.Restoration of turbulence-degraded images using anisotropic and nonlinear regularization[J].Journal of Astronautics,2004,25(1):5-12.]
[5] 洪汉玉,张天序,余国亮.基于Poisson模型的湍流退化图像多帧迭代复原算法[J].宇航学报,2004,25(6):649-654.[Hong Han-yu,Zhang Tian-xu,Yu Guo-liang.Iterative multi-frame restoration algorithm of turbulence-degraded images based on passion model[J].Journal of Astronautics,2004,25(6):649-654.]
[6] Fergus R,Singh B,Hertzmann A,et al.Removing camera shake from a single photograph[J].Acm Transactions on Graphics,2006,25(3):787-794.
[7] Krishnan D,Fergus R.Fast image deconvolution using hyper-Laplacian priors[C].International Conference on Neural Information Processing Systems,Brussels,Belgium,Sep 11-14,2011.
[8] Ohkoshi K,Goto T,Hirano S,et al.Blind image restoration based on total variation regularization of blurred images[C].The 1st IEEE Global Conference on Consumer Electronics,Tokyo,Japan,Oct 2-5,2012.
[9] Pan J,Sun D,Pfister H,et al.Blind image deblurring using dark channel prior[C].Computer Vision and Pattern Reco-gnition,Las Vegas,USA,Jun 27-30,2016.
[10] Krishnan D,Fergus R.Fast image deconvolution using hyper-Laplacian priors[C].International Conference on Neural Information Processing Systems,Vancouver,Canada,Dec 7-10,2009.
[11] Krishnan D,Tay T,Fergus R.Blind deconvolution using a normalized sparsity measure[C].Computer Vision and Pattern Recognition,Colorado Springs,USA,Jun 20-25,2011.
[12] Pan J,Su Z.Fast L0-regularized Kernel estimation for robust motion deblurring[J].IEEE Signal Processing Letters,2013,20(9):841-844.
[13] Levin A,Weiss Y,Durand F,et al.Understanding and evaluating blind deconvolution algorithms[C].Computer Vision and Pattern Recognition,Miami,USA,Jun 20-25,2009.