Imaging Complex Targets through a Scattering Medium Based on Adaptive Encoding

Round 1

Reviewer 1 Report

The authors proposed a deep learning-based method to reconstruct objects from the speckle pattern. In particular, the complicated human face can also be reconstructed by introducing the adaptive encoding. The following technical concerns should be properly addressed before it can be accepted for publication.

1.     Line 129, what is the meaning of “a larger range of information”.

2.     In section 3.2, the training groups of dual-characters and human faces are 7500 and 250， respectively. The feature of human face is much more complicated compared to that of dual-characters, but the authors use much fewer data to train the network. This is confusing.

3.     The experimental results shown in Figs.4 and 5 are the results with or without adaptive encoding are unclear. The authors should clarify this. 

4.     The adaptive encoding should be explained with more details. Also, the implement of the adaptive encoding in the experiment is unclear. It is loaded by the projector? The projector is used to display the ground truth. 

Author Response

Please see the attachment

Author Response File: Author Response.pdf

Reviewer 2 Report

1. What is the the abbreviation of GAN network? It is better for reading if mentioned its full term at their first appearance.

2. Line 45 says " At present, it is difficult to reconstruct complex targets with better quality even with the use of machine learning algorithms.". It is not very clear why it is difficult, how better the quality of reconstruction, and what improvement the machine learning can provide. Quantitative information are desired for others to understand the issue here. In other words, what are the goals of the manuscript? What are the limitations the methods break?

3. The newly proposed AESINet has better SNR. How about the time consumption of this method? 

4. From Fig. 4 and on are just the step-bystep results from Fig. 1. It looks sufficient. I just suggest the authors to more summarize the figure of merit of the proposed methods. 

Author Response

Please see the attachment

Author Response File: Author Response.pdf

Round 2

Reviewer 1 Report

The manuscript is now suitable for publication.