Ptychographic Algorithm Based On Sparse Representation

Phase retrieval is an issue that tries to recover an image from its Fourier magnitude measurements.Ptychography refers to reconstructing complex amplitude information of the image by using multiple diffraction patterns.The overlap ratio is the main factor affecting the imaging quality.Redundant information is greatly reduced under the scenario that the overlap ratio is relatively low,and the classical ptychographic phase retrieval algorithm cannot reconstruct the image effectively.Therefore,how to use the suitable image priors to improve the reconstruction quality of the image is the key issue.This paper focuses on ptychographic phase retrieval algorithms,the specific contents are as follows:Firstly,the sparsity of image over the discrete cosine transform is introduced into the ptychographic phase retrieval optimization problem.The paper proposes a ptychographic algorithm based on the sparsity of image over the discrete cosine transform.Through experiment contrast,show the effectiveness of the proposed algorithm.Secondly,the total variation can effectively reconstruct image edge and contour.Proposing the ptychographic algorithm based on the total variation.The algorithm incorporates the total variation regularization into the ptychographic phase retrieval problem,which based on amplitude constraints.Experimental results indicate that the performance of the proposed algorithm outperforms the traditional ptychographic algorithm.Moreover,it is robust to noise.Finally,the dual-tree complex wavelet can effectively keep image texture information.Proposing the ptychographic algorithm based on the dual-tree complex wavelet.The algorithm utilizes the Fourier magnitude measurements to construct a data fidelity term,and combines the sparse induced regularization term of the sparse representation of the image over the dual-tree complex wavelet transform to formulate a ptychographic phase retrieval optimization problem.Moreover,alternating direction method of multipliers is utilized for solving the corresponding optimization problem.The experimental results show that the proposed algorithm can obtain reconstructed image with high quality.