Your Location:
Home >
Browse articles >
Compressed wavefront sensing based on deep neural network for atmospheric turbulence
Intelligent Imaging | Updated:2023-06-12
    • Compressed wavefront sensing based on deep neural network for atmospheric turbulence

    • HUA Sheng-xiao

      13 ,  

      HU Qi-li

      2 ,  

      FENG Jia-hao

      13 ,  

      JIANG Lü

      13 ,  

      YANG Yan-yan

      13 ,  

      WU Jing-jing

      13 ,  

      YU Lin

      13 ,  

      HU Li-fa

      13 ,  
    • Chinese Journal of Liquid Crystals and Displays   Vol. 38, Issue 6, Pages: 789-797(2023)
    • DOI:10.37188/CJLCD.2023-0011    

      CLC: O439
    • Received:11 January 2023

      Revised:09 February 2023

      Published:05 June 2023

    Scan QR Code

  • Cite this article

    PDF

  • HUA Sheng-xiao, HU Qi-li, FENG Jia-hao, et al. Compressed wavefront sensing based on deep neural network for atmospheric turbulence[J]. Chinese journal of liquid crystals and displays, 2023, 38(6): 789-797. DOI: 10.37188/CJLCD.2023-0011.

  •  
  •  
    Sections

    Abstract

    When the compressive sensoring is used in wavefront measurement, classic methods of slopes’ restoration has a relatively low precision, which make it difficult to measure the atmospheric turbulence wavefront. In the paper, a deep neural network is presented to improve the slopes’ restoration precision. The traditional compressive sensing technology does not take into account the relatively small slopes, which increases the wavefront measurement errors. To measure the complex wavefront induced by atmospheric turbulence with a high speed, the paper presents an improved deep neural network to restore the slopes from sparse ones with high precision, which improves the precision of wavefront reconstruction. When the compression ratio is ranged from 0.1 to 0.9, the wavefront error PV (Peak to valley) of the compressed wavefront detection algorithm based on depth neural network (DNNCWS) proposed in this paper is better than 0.014 μm, and the running time of the algorithm is 4.4 ms. In the case of low signal-to-noise ratio, the residual wavefront PV is better than 0.011 μm. In addition, the simulation results indicate that it has good anti-noise performance. The DNNCWS improves the detection accuracy of compressive sensing and overcomes the problem of low accuracy for complex aberration induced by atmospheric turbulence. It can also be used in other adaptive optical applications, such as laser communication and retinal imaging.

    transl

    Keywords

    compressed wavefront sensing; adaptive optics; atmospheric turbulence

    transl

    1 引言

    在使用地基光学望远镜进行天文观测时,人们利用自适应光学系统(Adaptive optics system, AOS)补偿大气湍流造成的像差

    1-7。夏克哈特曼波前探测器(Shack-Hartmann wavefront sensor, SHWFS)是AOS中最重要的单元之一,它主要包括微透镜阵列和CCD。在天文观测中,自然导星或人造导星一般比较暗弱,同时大气湍流实时变化,大气湍流比较强时会导致SHWFS获得的部分光斑太弱,对这些光斑的质心进行计算时误差较大,导致波前重构精度低。同时,天文观测等应用要求SHWFS能快速对大气湍流的波前进行测量。虽然减少微透镜的数量和使用像素数较少的CCD可以提高波前测量速度,但会导致波前测量的空间分辨率降低,这意味着波前测量精度也会降低。
    transl

    近年来,人们将压缩感知技术引入波前测量,以少的斜率测量来提高波前测量速度。2014年,Polans提出了压缩波前探测,他利用Zernike多项式对斜率进行稀疏化并对由几个低阶Zernike模式组成的简单波前像差进行了仿真验证

    8。理论上它只需要较少数量的微透镜,因此可以提高波前的探测速度。Howland等人提出了一种基于压缩感知单像素相机的波前探测器,用于静态的暗弱目标的波前测量9。它将随机二进制图案应用于高分辨率空间光调制器,从10 000个投影中可以获得高质量的256×256波前。2016年,Eddy等人基于Polans的斜率稀疏化方法,使用压缩波前探测方法测量自由曲面光学元件的面形10。2022年,Ke等人使用压缩波前探测进行了波前校正实验11。然而,在前述已报道的稀疏斜率重构方法中,具有较小值的斜率通常被忽略为0,这明显增加了小压缩比时的波前重构误差,因此,他们主要用低阶Zernike模式组成的简单波前进行重构和验证,而无法测量大气湍流引起的复杂波前畸变。大气湍流引起的像差非常复杂,导致斜率分布范围广,而小的斜率值也可能对重构复杂波前的高频成分有贡献,直接视为0会导致波前中的部分高频信息丢失。如果斜率恢复的精度低会导致波前重构的精度相应降低,因此,已有的压缩波前探测算法很难应用于地基大口径光学望远镜的自适应光学系统中。
    transl

    为了解决该问题,我们提出了深度神经网络来提高稀疏斜率的恢复精度。传统的算法可以从光点阵计算斜率并重构波前,但针对暗弱空间目标探测时,湍流扰动引起的光点亮度起伏会导致部分光点信噪比差,甚至不能用于波前重构,导致重构精度降低。近年来,人们利用深度学习技术直接从光点阵预测波前且具有较高的波前重构精度,但它们的算法严重依赖于硬件和数据训练,计算时间通常高达几十毫秒

    12-15,外场环境中的光点阵和对应波前的数据集很难同时准确获得,使其目前还不能用于外场的大气湍流校正。在本文所提出的深度神经网络中,输入和输出数据是斜率而不是具有大网格数的光斑和波前图像,并且即使对于具有30×30微透镜的SHWFS,斜率数据个数也不大于1 800。因此,我们设计的深度神经网络可以在其快速恢复速度和高波前重构精度之间取得平衡,通过高精度地恢复斜率提高压缩波前探测算法中波前重构精度。
    transl

    2 方法

    仿真过程如图1所示。从光点阵列图像计算沿xy方向的原始斜率

    4,然后利用离散余弦变换(DCT)矩阵生成稀疏斜率;沿xy方向的稀疏斜率被用作深度神经网络(DNN)的输入参数。作为比较,图1中红色箭头还给出了压缩波前探测的经典恢复方法。沿xy方向的恢复斜率为输出结果。最后,可以使用Zernike模式法重构波前。
    transl

    fig

    图1  压缩波前探测的波前重构过程

    Fig.1  Wavefront reconstruction process of compressive wavefront sensor

    icon Download:  Full-size image | High-res image | Low-res image

    2.1 SHWFS波前重构

    SHWFS通过测量光斑阵列并计算其沿xy方向的偏移,根据斜率重构波前。在SHWFS中,波前通过微透镜阵列形成光点阵列。每个子孔径上的局部波前的平均倾斜和尖端与光点的偏移成比例。偏移可以通过比较从参考波前获得的光点的质心与畸变波前的质心之间的差异来计算。通常用传统的重心(CoG)算法计算:

    transl

    xc=i=imini=imaxj=jminj=jmaxIi,j×ii=imini=imaxj=jminj=jmaxIi,j×c (1)
    yc=i=imini=imaxj=jminj=jmaxIi,j×ji=imini=imaxj=jminj=jmaxIi,j×c (2)

    其中Iij)是第i行和第j列中像素的光强。然后,根据光点的偏移xy计算波前斜率。定义φxy)为坐标(xy)处的大气湍流波前相位,其斜率分别为xy方向上的SxSy。它们之间的关系如下:

    transl

    Sx=dφx,ydxΔxf (3)
    Sy=dφx,ydyΔyf (4)

    其中f是透镜的焦距。计算斜率后,可以通过经典波前重构方法重构最终波前W,如式(5)所示

    416-17
    transl

    W=iaiZi (5)

    其中:Zi是第i个Zernike模式,ai是第i个Zernike模式的系数。

    transl

    2.2 斜率稀疏和恢复

    定义ΨRM×N作为MN的稀疏基,这意味着空间中的每个向量都可以表示为这组基向量的线性组合。然后,斜率信号SxSy可以通过稀疏基进行稀疏化,如式(6)所示:

    transl

    Sx=Ψθx,Sy=Ψθy (6)

    其中稀疏向量θxRNθyRN分别是SxSy的稀疏表示。离散余弦变换(DCT)矩阵被用作稀疏矩阵。

    transl

    设计一个M×N探测矩阵来测量SxSy并获得斜率测量值S'xS'y。我们选择高斯矩阵作为探测矩阵:

    transl

    Sx'=ΦSx,Sy'=ΦSy . (7)

    根据方程(5)和(6),S'xS'y可以表示如下:

    transl

    Sx'=ΦΨθx=ASx,Sy'=ΦΨθy=ASy (8)

    其中A=ΦΨ

    transl

    矩阵Φ满足限制等距性质(RIP)标准

    18-20
    transl

    1-εS2ΦS21+εS2 . (9)

    因为MN小得多,所以S'xS'y的值通过求解最小0范数来得到。而0范数问题是一个NP难问题,但Candes、Tao和Donoho等人已证明,当测量矩阵Φ满足约束等距性质(RIP)时,l0约束优化问题可转化为l1约束的凸优化问题:

    transl

    minΨSx1 subject to ASx=ΦΨSx=Sx' (10)
    minΨSy1 subject  to ASy=ΦΨSy=Sy' . (11)

    恢复斜率S'xS'y与原始斜率之间的差异应足够小才得以获得较高的波前测量精度。该算法用于重构波前斜率的传统压缩感知算法,可以命名为基于高斯的压缩波前探测算法(GCWS)。本文提出了一种基于深度神经网络的压缩波前探测算法(DNNCWS)以提高恢复波前斜率的准确性。由于大气湍流的影响,用于地基光学望远镜的自适应光学系统时,在夏克哈特曼波前探测器探测得到的一帧光点图像中,各个光点的光强有起伏,部分光点信噪比很差,用于波前重构时甚至会导致误差的增加。另一方面,由于波前探测器CCD量子效率的限制,微透镜数不能太多,以免探测暗弱目标的能力降低。本文针对快速恢复稀疏斜率的情况,设计了9层神经网络结构,其结构如图2所示。第一层为conv1层,卷积核大小为3×3,步长为1,padding是1,并且使用Batch Normalization进行归一化。第二层到第六层为BasicBlock层,即Resnet中使用到的残差模块,该层包括两个conv层,卷积核大小为3×3,步长为1,padding为1,将ReLu函数作为激活函数并进行归一化。第七、第八层为DoubleConv模块,由两个conv层构成,卷积核大小为3×3,步长为1,padding是1,使用Batch Normalization进行归一化并将ReLu函数作为激活函数。第九层为OutConv层,卷积核大小为1×1,不包含Batch Normalization与激活函数ReLu。对于稀疏化的波前斜率,该网络可以将稀疏化的波前斜率在更短时间内以更高的精度恢复原始斜率,进行高精度的波前重构。共生成了30 000组波前和斜率数据用于训练。在训练过程中,使用具有不同压缩比的斜率。在获得最优模型后,可以实现任意稀疏波前斜率的高精度恢复。

    transl

    fig

    图2  深度神经网络结构

    Fig.2  Deep neural network structure

    icon Download:  Full-size image | High-res image | Low-res image

    3 结果与讨论

    本文在J波段进行了模拟,望远镜孔径D为3.6 m,r0范围为0.33~0.42 m。SHWFS中的微透镜数量为16×16,每个子区域的像素数为6×6。压缩比r表示所有斜率中有效斜率的百分比。

    transl

    3.1 DNNCWS的波前重构精度

    对于相同的压缩比r=0.7,分别使用DNNCWS和GCWS重构波前斜率及其波前。图3给出了xy方向上波前斜率及其重构误差。图3(a)为x方向上的原始斜率,其值范围为-0.7~0.8 rad;图3(f)为y方向上的原始斜率,其值为-0.9~1.1 rad。

    transl

    fig

    图3  (a)原始波前的x方向斜率;(b)GCWS恢复的x方向斜率;(c)GCWS恢复的x方向斜率的误差;(d)DNNCWS恢复的x方向斜率;(e)DNNCWS恢复的x方向斜率的误差;(f)原始波前的y方向斜率;(g)GCWS恢复的y方向斜率;(h)由GCWS恢复的y方向斜率的误差;(i)由DNNCWS恢复的y方向斜率;(j)由DNNCWS恢复的y方向斜率误差。

    Fig.3  (a) X-direction slopes of the original wavefront; (b) X-direction slopes recovered by GCWS; (c) Error of the x-direction slopes recovered by GCWS; (d) X-direction slopes recovered by DNNCWS; (e) X-direction slopes error recovered by DNNCWS; (f) Y-direction slopes of the original wavefront; (g) Y-direction slopes recovered by GCWS; (h) Error of the y-direction slopes recovered by GCWS; (i) Y-direction slopes recovered by DNNCWS; (j) Y-direction slopes error recovered by DNNCWS.

    icon Download:  Full-size image | High-res image | Low-res image

    图3(b)和(g)给出了GCWS恢复的xy方向的斜率。图3(c)和(h)分别给出了GCWS恢复的xy方向上斜率的误差,其PV为0.5 rad,RMS为0.03 rad。图3(d)和(i)分别给出了DNNCWS恢复的xy方向的斜率。图3(e)和(f)给出了相应的误差,其PV为0.04 rad,RMS值为0.003 rad。可以看出,DNNCWS算法的精度比常规压缩波前探测算法高了一个量级。

    transl

    基于所获得的斜率进行波前重构,得到了如图4所示的重构波前。所用电脑CPU为AMD Ryzen7,内存8 GB,从斜率到波前的重构时间为4.4 ms。图4(a)、(b)和(c)分别给出了原始波前、DNNCWS重构波前和残差波前。图4(d)和(e)分别给出了GCWS重构波前和残差波前。图4(a)中原始波前的峰谷值(PV)为0.464 μm,均方根值(RMS)为0.063 5 μm。图4(b)中的波前PV为0.460 μm,RMS为0.063 3 μm。图4(d)中的波前PV为0.379 μm,RMS为0.061 5 μm。图4(c)中DNNCWS的残差波前的RMS为0.001 4 μm,远小于图4(e)中GCWS的残差波前RMS 0.006 μm。DNNCWS获得的结果的精度远高于GCWS,这是由于DNNCWS方法恢复斜率具有更高的精度,因而降低了波前重构误差。

    transl

    fig

    图4  (a)原始波前;(b)DNNCWS算法恢复的斜率重构的波前;(c)DNNCWS算法的残差波前;(d)GCWS算法恢复的斜率重构的波前;(e)GCWS算法的残差波前。

    Fig.4  (a) Original wavefront; (b) Reconstructed wavefront with DNNCWS; (c) Residual wavefront of DNNCWS; (d) Reconstructed wavefront with GCWS; (e) Residual wavefront of GCWS.

    icon Download:  Full-size image | High-res image | Low-res image

    为了检验算法的稳定性,我们选择了30组数据进行了测试。图5显示了使用两种方法从30组随机斜率中恢复的残差波前的PV和RMS。可以看出,DNNCWS比GCWS具有显著的稳定性和高精度。DNNCWS算法的残差波前的RMS值约为0.001 μm,变化幅度较小。

    transl

    fig

    图5  30组斜率的残差波前的PV和RMS误差的比较。(a)误差的PV值;(b)误差的RMS值。

    Fig.5  Comparison of PV and RMS errors of residual wavefront for 30 sets of slopes. (a) PV values of errors; (b) RMS values of errors.

    icon Download:  Full-size image | High-res image | Low-res image

    3.2 不同的斜率压缩比对波前重构精度的影响

    图6显示了两种算法在不同压缩比下恢复的残差波前的PV和RMS的比较。基于DNNCWS重构的残差波前的PV范围为0.005~0.014 μm,RMS范围为0.000 7~0.002 μm;基于GCWS重构的残差波前PV范围为0.07~0.4 μm,RMS范围为0.004~0.066 μm。结果表明,DNNCWS的波前测量精度远优于GCWS。与GCWS相似,随着斜率压缩比的增加,DNNCWS重构的残差波前的RMS从0.002 μm降低到0.000 7 μm。在所有情况下,DNNCWS的残差波前的RMS均在0.001 μm左右。DNNCWS使用深度神经网络来提高从稀疏波前斜率恢复的波前斜率的准确性,因此重构的波前效果更好。同时,对于相同的波前测量精度,DNNCWS可以使用较少的微透镜进行测量。

    transl

    fig

    图6  不同压缩比下的残差波前比较。(a)残差波前PV的比较;(b)残差波前RMS的比较。

    Fig.6  Comparison of residual wavefront under different compression ratios. (a) Comparison of PV of residual wavefront; (b) Comparison of RMS of residual wavefront.

    icon Download:  Full-size image | High-res image | Low-res image

    3.3 暗弱星等下压缩波前探测的精度

    对于比较暗弱的11星等目标,在信噪比比较低(SNR=8.11 dB),压缩比r=0.7的情况下,图7(a)、(b)和(c)分别为原始波前、DNNCWS算法恢复的斜率重构的波前与残差波前,图7(d)和(e)分别为GCWS算法恢复的斜率重构的波前与残差波前。图7(a)中原始波前的PV值为0.448 μm,RMS值为0.064 1 μm;图7(b)中DNNCWS算法恢复的波前PV值为0.458 μm,RMS值为0.062 6 μm;图7(d)中GCWS算法恢复的波前PV值为0.407 μm,RMS值为0.062 0 μm。图7(c)中基于DNNCWS重构的波前残差RMS值为0.000 9 μm,图7(e)中基于GCWS重构的波前残差的RMS值为0.004 μm。可以看出,DNNCWS恢复的波前比GCWS的结果更接近原始波前。在低信噪比条件下,基于DNNCWS的波前重构精度显著高于GCWS。

    transl

    fig

    图7  (a)原始波前;(b)DNNCWS算法恢复的斜率重构的波前;(c)DNNCWS算法恢复的斜率重构的波前的残差波前;(d)GCWS算法恢复的斜率重构的波前;(e)GCWS算法恢复的斜率重构的波前的残差波前。

    Fig.7  (a) Original wavefront; (b) Reconstructed wavefront with DNNCWS; (c) Residual wavefront of DNNCWS; (d) Reconstructed wavefront with GCWS; (e) Residual wavefront of GCWS.

    icon Download:  Full-size image | High-res image | Low-res image

    图8(a)和(b)分别为在低星等情况下30组随机数据所得到的波前残差的RMS值与PV值的对比。由图8(a)和(b)可知,基于DNNCWS算法重构的波前残差PV值在0.004~0.013 μm范围内,RMS值在0.000 7~0.001 9 μm范围内;而GCWS算法对应的波前残差的PV值在0.066~0.198 μm范围内,RMS值在0.004~0.008 μm范围内。显然,DNNCWS比GCWS具有更高的精度和更好的稳定性。

    transl

    fig

    图8  30组斜率的波前误差比较。(a)残差波前PV的比较;(b)残差波前RMS的比较。

    Fig.8  Wavefront error comparison of 30 sets of slopes. (a) PV of residual wavefront; (b) RMS of residual wavefront.

    icon Download:  Full-size image | High-res image | Low-res image

    图9(a)和(b)分别显示了重构波前的RMS和PV在不同星等下重构波前的残差对比。星等数值越高,目标光源越暗,信噪比也会越低。5、7、9、11和13星等对应的信噪比SNR分别为49.74,32.2,15.08,8.11,6.76 dB。结果表明,基于DNNCWS重构的残差波前PV在0.005~0.011 μm之间,RMS在0.000 9~0.001 7 μm之间。作为比较,基于GCWS重构的残差波前PV范围为0.06~0.11 μm,RMS范围为0.004~0.005 μm。DNNCWS的波前重构精度较高,不同的幅度对其波前测量精度的影响较小。

    transl

    fig

    图9  不同星等下的残差波前。(a)残差波前PV的比较;(b)残差波前RMS的比较。

    Fig.9  Residual wavefront under different magnitude of star. (a) RMS of residual wavefront; (b) PV of residual wavefront.

    icon Download:  Full-size image | High-res image | Low-res image

    4 结论

    本文提出DNNCWS的方法,采用深度神经网络从稀疏的斜率恢复斜率,提高了压缩波前的检测精度。与已有的压缩波前探测方法相比,该方法克服了测量复杂像差精度低的问题,可以用于大气湍流引起的复杂波前测量。此外,DNNCWS可以在不同压缩比和低信噪比的条件下以高精度和稳定性重构波前。在压缩比从0.1~0.9变化情况下,DNNCWS的PV范围为0.005~0.014 μm,残差波前的RMS均在0.001 μm左右,算法的运行时间为4.4 ms。在暗弱星等情况下,即低信噪比条件下,DNNCWS的残差波前PV在0.005~0.011 μm之间,RMS在0.000 9~0.0017 μm之间。本文所提出的方法在使用自适应光学的其他光学成像系统中也将有较好的应用。

    transl

    参考文献

    1

    ANGEL J R P. Ground-based imaging of extrasolar planets using adaptive optics [J]. Nature19943686468): 203-207. doi: 10.1038/368203a0 [Baidu Scholar] 

    2

    HU L FXUAN LLIU Y Jet al. Phase-only liquid-crystal spatial light modulator for wavefront correction with high precision [J]. Optics Express20041226): 6403-6409. doi: 10.1364/opex.12.006403 [Baidu Scholar] 

    3

    赵子云顾虎马文超自适应光学系统误差分析与参数优化研究[J].液晶与显示2021365):663-672 doi: 10.37188/CJLCD.2020-0356 [Baidu Scholar] 

    ZHAO Z YGU HMA W Cet al. Error budget and parameters optimization of adaptive optics system [J]. Chinese Journal of Liquid Crystals and Displays2021365): 663-672. (in Chinese). doi: 10.37188/CJLCD.2020-0356 [Baidu Scholar] 

    4

    GU HZHAO Z YZHANG Z Get al. High-precision wavefront reconstruction from Shack-Hartmann wavefront sensor data by a deep convolutional neural network [J]. Measurement Science and Technology2021328): 085101. doi: 10.1088/1361-6501/abf708 [Baidu Scholar] 

    5

    DUNCAN J LZHANG Y HGANDHI Jet al. High-resolution imaging with adaptive optics in patients with inherited retinal degeneration [J]. Investigative Ophthalmology & Visual Science2007487): 3283-3291. doi: 10.1167/iovs.06-1422 [Baidu Scholar] 

    6

    POYNEER L AGAVEL D TBRASE J M. Fast wave-front reconstruction in large adaptive optics systems with use of the Fourier transform [J]. Journal of the Optical Society of America20021910): 2100-2111. doi: 10.1364/josaa.19.002100 [Baidu Scholar] 

    7

    陈浩宣丽胡立发望远镜的紧凑型闭环液晶自适应光学系统设计[J].液晶与显示2010253):379-385 doi: 10.3969/j.issn.1007-2780.2010.03.017 [Baidu Scholar] 

    CHEN HXUAN LHU L Fet al. Design on compact type closed-loop liquid crystal adaptive optical system for telescope [J]. Chinese Journal of Liquid Crystals and Displays2010253): 379-385. (in Chinese). doi: 10.3969/j.issn.1007-2780.2010.03.017 [Baidu Scholar] 

    8

    POLANS JMCNABB R PIZATT J Aet al. Compressed wavefront sensing [J]. Optics Letters2014395): 1189-1192. doi: 10.1364/ol.39.001189 [Baidu Scholar] 

    9

    HOWLAND G ALUM D JHOWELL J C. Compressive wavefront sensing with weak value [J]. Optics Express20142216): 18870-18880. doi: 10.1364/oe.22.018870 [Baidu Scholar] 

    10

    TIK E C MWANG XGUO N Qet al. Investigation on compressed wavefront sensing in freeform surface measurements [C]. 2016 4th International Conference on Photonics, Optics and Laser Technology. RomeIEEE20161-6. doi: 10.5220/0005743901520157 [Baidu Scholar] 

    11

    KE X ZWU J LHAO J X. Distorted wavefront reconstruction based on compressed sensing [J]. Applied Physics B2022128107. doi: 10.1007/s00340-022-07827-6 [Baidu Scholar] 

    12

    NISHIZAKI YVALDIVIA MHORISAKI Ret al. Deep learning wavefront sensing [J]. Optics Express2019271): 240-251. doi: 10.1364/oe.27.000240 [Baidu Scholar] 

    13

    GUO H YXU Y JLI Qet al. Improved machine learning approach for wavefront sensing [J]. Sensors20191916): 3533. doi: 10.3390/s19163533 [Baidu Scholar] 

    14

    VISHNIAKOU ISEELIG J D. Wavefront correction for adaptive optics with reflected light and deep neural networks [J]. Optics Express20202810): 15459-15471. doi: 10.1364/oe.392794 [Baidu Scholar] 

    15

    JIA PMA M YCAI D Met al. Compressive Shack-Hartmann wavefront sensor based on deep neural networks [J]. Monthly Notices of the Royal Astronomical Society20215033): 3194-3203. doi: 10.1093/mnras/staa4045 [Baidu Scholar] 

    16

    SOUTHWELL W H. Wave-front estimation from wave-front slope measurements [J]. Journal of the Optical Society of America1980708): 998-1006. doi: 10.1364/josa.70.000998 [Baidu Scholar] 

    17

    FRIED D L. Least-square fitting a wave-front distortion estimate to an array of phase-difference measurements [J]. Journal of the Optical Society of America1977673): 370-375. doi: 10.1364/josa.67.000370 [Baidu Scholar] 

    18

    BARANIUK RDAVENPORT MDEVORE Ret al. A simple proof of the restricted isometry property for random matrices [J]. Constructive Approximation2008283): 253-263. doi: 10.1007/s00365-007-9003-x [Baidu Scholar] 

    19

    CANDÈS E J. The restricted isometry property and its implications for compressed sensing [J]. Comptes Rendus Mathematique20083469/10): 589-592. doi: 10.1016/j.crma.2008.03.014 [Baidu Scholar] 

    20

    DANDES E JTAO T. Near-optimal signal recovery from random projections: Universal encoding strategies [J]. IEEE Transactions on Information Theory20065212): 5406-5425. doi: 10.1109/tit.2006.885507 [Baidu Scholar] 

    474

    Views

    162

    Downloads

    0

    CSCD

    Alert me when the article has been cited
    Submit
    Tools
    Download
    Export Citation
    Share
    Add to favorites
    Add to my album

    Related Articles

    Wavefront correction method based on P-U-net for pyramid wavefront detector
    High-precision turbulence wavefront reconstruction based on Transformer structure
    Research on improving the universality of adaptive optical retinal imaging system
    Design and performance research of high efficiency variable reluctance voice coil actuator

    Related Author

    Hua Sheng-xiao
    Jiang Lv
    HU Ming
    ZHANG Qi
    WANG Hongyan
    XU Xingyu
    HU Qili
    WU Jingjing

    Related Institution

    Key Laboratory of Electro-Optical Countermeasure Test & Evaluation Technology
    College of Science, Jiangnan University
    Jiangsu Provincial Research Center of Light Industry Optoelectronic Engineering and Technology
    School of Science, Jiangnan University
    Key laboratory of Electro-Optical Countermeasures Test and & Evaluation Technology
    0