Improved Error Reduction and Hybrid Input Output Algorithms for Phase Retrieval by including a Sparse Dictionary Learning-Based Inpainting Method

Figure 2

The KSVD algorithm consists of the sparse coding and dictionary update stages, where and represent the iteration and the number of dictionary atoms, respectively. Sparse coding stage: apply the pursuit algorithms to solve coefficient vectors for each signal while fixing . Dictionary update stage: the dictionary is updated column by column through the SVD. The two stages are iteratively performed until convergence. More details about this algorithm can be found in [30].