【作者】 高颖；

【导师】 陈书贞；

【作者基本信息】 燕山大学 ， 信息与通信工程， 2017， 硕士

【摘要】 同轴数字全息技术已被广泛应用于各个领域,但在传统的全息再现过程中,重建得到的图像质量会受到共轭像的严重影响。利用相位恢复算法可以借助测得的全息图强度值恢复得到丢失的相位信息,从而彻底消除共轭像的影响,本文主要围绕基于同轴数字全息的相位恢复重建算法展开研究,具体研究内容如下:首先,针对纯相位物体提出了一种基于同轴数字全息的抗泊松噪声相位恢复算法。该算法以负对数的泊松似然函数作为数据保真项,将二阶广义全变差(TGV)作为抗泊松噪声污染模型的正则项。实验结果表明,当测量值受到不同强度的泊松噪声干扰时,该算法可以消除共轭像的影响,使得图像实现精确重构。其次,提出了一种基于二阶广义全变差和全变差(TV)混合正则化的同轴数字全息重建算法,该算法利用TGV正则项和TV正则项分别约束重构数据的空间幅值和相位分布,并将复图像表示为两种不同的形式,利用交替方向乘子法(ADMM)方法求解相应的优化问题。实验结果证明了提出算法的有效性和鲁棒性,另外,即使在较低的过采样率下,利用提出的算法对图像进行重构仍可以得到较好的效果。最后,在同轴数字全息的基础上,提出了一种基于二阶广义全变差正则化的重构算法。针对实值和纯相位物体,该算法利用广义全变差正则项约束重构数据;针对复图像,利用广义全变差正则项分别约束重构数据的空间幅值和相位分布。结果表明将TGV作为正则项时可以去除阶梯效应,并能较大程度地保留图像的边缘和细节信息,使得重建得到的图像更加真实,从而证明了提出算法的有效性。更多还原

【Abstract】 In-line digital holography has been widely used in various fields,but in the process of traditional hologram reconstruction,the quality of reconstructed image suffers from the superposition of twin-image contamination.The phase retrieval algorithm is used to recover the lost phase information with the help of the measured hologram intensity,which can eliminate the influence of twin-image image completely.In this paper,we focus on the research of phase retrieval reconstruction algorithm,which based on in-line digital holography.The main contents are as follows:Firstly,a resistance to poisson noise pollution phase retrieval algorithm based on in-line digital holography is proposed for pure phase object.The negative-log poisson likehood function is used for data fidelity term and second order total generaliazed variation(TGV)is used for the regularization term of the model.Experimental results show that the proposed algorithm can eliminate the interference of the twin-image in the case of Poisson noise with different intensities,making the image reconstruction accurately.Secondly,a new in-line digital holography reconstruction algorithm is proposed,which based on the second order total generaliazed variation and total variation(TV)hybrid regularization.By using TGV regularization and TV regularization in the proposed algorithm,which can constraint spatial amplitude and phase distribution of the reconstructed data separately.Moreover,the complex image can be represented in two different forms.The alternating direction method of multipliers(ADMM)is utilized for solving the corresponding optimization problem.Experimental results demonstrate the effectiveness and robustness of the proposed algorithm,in addition,even at a low oversampling rate,the proposed algorithm can achieve better image reconstruction quality.Finally,on the basis of in-line digital holography,we proposed a new image reconstruction algorithm based on the second order total generaliazed variation.For the real and pure phase object,the second order total generaliazed variation regularization can constrain the reconstructed data.For the complex-valued image,the second order total generaliazed variation regularization can constrain the spatial amplitude and phase distributions of the reconstructed data separately.Experimental results show that TGV can eliminate the staircase effect,and can preserve the edge and detail information of the image as much as possible.The reconstructed image is more real when the second order total generaliazed variation is used for the regularization term.Thus,reflect the effectiveness of the algorithm.更多还原