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

<div>Taking the text object of “NCTU” as an example, two performance metrics were evaluated in the five PR algorithms. (a) RSE and (b) SSIM. Since KSVD is executed at each iteration cycle and results in spikes, the perturbation enables the algorithm to jump out of the stagnant local optimum.</div>

International Journal of Optics

fig9

Figure 9

Figure 9: Improved Error Reduction and Hybrid Input Output Algorithms for Phase Retrieval by including a Sparse Dictionary Learning-Based Inpainting Method 

Figure 9 | Improved Error Reduction and Hybrid Input Output Algorithms for Phase Retrieval by including a Sparse Dictionary Learning-Based Inpainting Method 