Next Article in Journal
IMPAβ: Incremental Modal Pushover Analysis for Bridges
Next Article in Special Issue
Providing a Visual Understanding of Holography Through Phase Space Representations
Previous Article in Journal
Accelerating Super-Resolution and Visual Task Analysis in Medical Images
Previous Article in Special Issue
Partially Overlapping Printing with Digital Filter to Improve Quality of Volume Hologram Printer – Numerical Simulation –
 
 
Article
Peer-Review Record

Optimization of Phase-Only Computer-Generated Holograms Based on the Gradient Descent Method

Appl. Sci. 2020, 10(12), 4283; https://doi.org/10.3390/app10124283
by Shujian Liu * and Yasuhiro Takaki
Reviewer 1: Anonymous
Reviewer 2: Anonymous
Appl. Sci. 2020, 10(12), 4283; https://doi.org/10.3390/app10124283
Submission received: 27 May 2020 / Revised: 17 June 2020 / Accepted: 17 June 2020 / Published: 22 June 2020
(This article belongs to the Special Issue Practical Computer-Generated Hologram for 3D Display)

Round 1

Reviewer 1 Report

Liu et al., Optimization of phase-only computer-generated holograms based on the gradient descent method

 

The authors propose an optimization of the phase Fourier hologram using gradient descent technique. Unique point is that they define the error in intensity rather than in amplitude like standard techniques. At each iteration, the hologram phase is updated using the gradient of the intensity error. The optimum update rate is also derived. Simulations and experimental results are presented showing lower error than conventional GS algorithm.

 

The GS algorithm is a well-known technique equivalent to the gradient descent algorithm defined for the amplitude error as the authors mentioned. The authors argue that this is the first report on its variation for the intensity error. I am not 100% sure as I did not search prior works extensively. But it would be novel and interesting work if it is true. I think this paper can be considered for publication in Applied Sciences after some revisions. Followings are my comments.

 

  1. Currently, the proposed technique is only compared with the GS algorithm. It is recommended to compare it with other competing algorithms as well like a genetic algorithm.

 

  1. In line 124 to 127, authors said that intensity variation corresponding to black region is larger for the proposed method as shown in Figure 8. However the reason for this is not mentioned.

 

  1. For MSE using GS algorithm, the error curve seems to be converging and not improving much after few iterations. However using proposed method the curve does not seem to be converging even at 100 iterations. So why authors used 100 iteration if they can improve MSE more.

 

  1. Experimental results should be compared to the results using GS algorithm for better comparison as authors did for numerical simulation in Figure 6.

 

  1. It would be good to present the evolution of the optimized error reduction rate in Fig. 9.

 

  1. In Fig. 7, did you use the same MSE definition (MSE measured in intensity) for GS and the proposed methods?

 

  1. In Fig. 10, where is the Fourier transform lens mentioned in the main text?

 

  1. Could you extend the proposed method to Fresnel regime?

 

  1. Figures 4 and 5 are not cited in the main text.

Author Response

Please see the attachment

Author Response File: Author Response.pdf

Reviewer 2 Report

Attached.

Comments for author File: Comments.pdf

Author Response

Please see the attachment

Author Response File: Author Response.pdf

Round 2

Reviewer 1 Report

The authors added new experimental results and revised the manuscript, addressing all the issues raised in the previous review. I think that the manuscript can now be considered for publication in Applied Sciences in its present form.

Back to TopTop