Fast Computational Ghost Imaging with Laser Array Modulation
Round 1
Reviewer 1 Report
The authors present a new idea on optimized ghost imaging using modulated laser arrays. The core idea is to use a computational ghost imaging scheme, where a spatial light modulator is used to create a scattering pattern and instead of changing the pattern, the illumination of this pattern is changed by using a different laser from the array that is located at a different position. The authors then imply a significant speed-up compared to standard ghost imaging as the lasers can be modulated more quickly than the SLM. The authors then go on and realize an experiment analogous to the approach they propose by investigating a single movable laser instead of a laser array, a stationary grond glass disk instead of a SLM and two image detectors instead of a single bucket detector.
In my opinion, the idea presented here might be of some interest, but is lacking currently as the idea is not presented in enough detail and there is little to no discussion with respect to other works or approaches existing in the literature to reach similar targets.
First, it is not entirely clear to he reader what the authors want to do in detail. Do the authors want to switch on one laser after another or will it be possible for several lasers to emit simultaneously? It seems like the authors intend to follow the first scenario, but it never becomes really clear. If so, as the scattering pattern imprinted on the SLM is essentially static, this is not too different from using a precomputed set of patterns or avoiding the SLM altogether by placing a different ground glass disk in front of every single VCSEL used. In this case, the authors should find some motivation for their approach. What would be the fastest realistic effective frame rate they can realize? I guess it would be limited by the temporal resolution of the bucket detector and by how quickly the signal on the detector becomes stable. If the authors indeed aim at the kHz range, they should also discuss other techniques that work in a similar manner. For example a setup using a single large laser spot and a digital micromirror device (DMD) instead of the VCSEL array would serve the same purpose. The individual mirrors on the DMD may then be switched such that only one of the micromirrors is active at any instant and only the light from this mirror arrives at the SLM. As frame rates of DMDs can easily reach 50 kHz or more, the authors should state clearly, why one should implement their array approach instead of a DMD.
Second, the manuscript shows some deviation between what is promised and what is delivered. The authors discuss ultrafast computational imaging with a VCSEL array and then show an analogy experiment with a single laser that is moved around, a phase pattern that is not computed, but measured explicitly and an image chip instead of a bucket detector. While analogy experiments are certainly fine, from reading the title and the abstact the reader might get the impression that the fast imaging scheme proposed by the authors is actually realized in this manuscript. The authors should state very clearly and prominently that this is not the case and clearly distinguish between what is suggested and what is actually realized.
In summary, the manuscript might be of interest to the ghost imaging community, but at current the manuscript lacks clarity and it is not clear whether the approach presented here really yields advantages compared to e.g. DMDs. Accordingly, at current I cannot recommend the present manuscript for publication at current. The authors should at least provide some clear discussion that shows that their approach is superior to other approaches at least for some experimental conditions or in some special setting.
Some minor points that should also be addressed:
- line 50: When the authors state that the SLM patterns should be updated "more than thousand times", they are quite unclear. More than thousand times per what?
- The manuscript contains plenty of typos. The authors may want to reread it. Onr example is the usage of corrected instead of collected in line 72.
- Figure 2 is really small and hard to read. The figure or at least the font should be enlarged.
Author Response
Dear reviewer 1,
I appreciate the reviewer 1 for many helpful and appropriate comments.
We carefully revised our manuscript in accordance with your comments.
Details of our reply is described the attached file.
Author Response File: 
Author Response.docx
Reviewer 2 Report
Dear Editor
This paper presents a new method for computational ghost imaging, where the amplitude modulation of different emitters of a laser array is used to refresh speckle pattern with the purpose of increasing the acquisition speed.
The idea is of a certain interest and it can be worth presenting it to the community interested in ghost imaging. Nonetheless, in my opinion, some improvement should be done before publication.
At page 3 is mentioned “”should allow a dramatic reduction in the CGI measurement time”. A more quantitative analysis is needed.
When quoting ref.13 in relation to Eq.1, one should mention also the earlier PRA 79, 053840 (2009)
A more quantitative analysis, for example in terms of signal to noise ratio of the achieved result (e.g. see Phys Rev. A 83, 063807 (2011)), should be done when discussing FIG.5, comparing with other highly performing GI techniques e.g. NGI, optimized differential ghost imaging (arXiv:1903.12630), …
More experimental details should be provided in section 4, as intensity of the laser, signal to noise ration of the achieved result, …
Furthermore, it would be useful comparing the new method with a traditional GI realized with the same set up.
References are limited and sometimes incorrect:
-When mentioned “the first GI proposal [1]” it should be noticed that the original proposal was of Klyshko (Belinskii A and Klyshko D 1994 Sov. Phys. JETP 78 259), while Ref.1 is the first experimental realisation.
-The fact that split thermal light suffices for GI was firstly suggested by Bennink R et al 2002 Phys. Rev. Lett. 89 113601 (I suggest adding this reference).
-Since Applied Science addresses a general audience I suggest quoting some review paper in the introduction, e.g. Journal of Optics, 18 (2016) 073002 and Nature Reviews Physics 1, 367.
-A few typos shoud be corrected, eg. missing lines between initial and final page in several references.
In general the paper should be better presented and motivated.
Author Response
Dear reviewer 2,
I appreciate the reviewer 2 for many helpful and appropriate comments.
We carefully revised our manuscript in accordance with your comments.
I send the attached document for our reply.
Yours sincerely,
Author Response File: Author Response.docx
Reviewer 3 Report
The authors present a 'proof of principle' scheme to decrease the speckle acquisition time required for ghost imaging. The manuscript is generally acceptable, however there are a few points that deserve some attention.
1) There is some potential for confusion regarding the laser array- for example, are all the lasers identical, meaning the same spectral output?
2) Line 74, the sentence beginning "Another data set of..." is confusing, because the prior sentences seem to state that the intensity patterns are all obtained during a measurement.
3) Lines 120-127, the paragraph is unclear because the focal length of the condenser lens was not specified. Similarly, the speckle size should depend on the condenser lens focal length and location relative to the image plane.
4) Line 151, the sentence "Bi is derived from the sum of all pixel values" is problematic because a single photo-detector will respond to the complex-valued field incident on it, but summing pixel values only considers the intensity, so it's not clear that these are equivalent.
5) Figure 4 shows a setup that is significantly different from that shown in Figure 1, but the authors do not comment on this.
6) Line 173, the authors state that targets with large transparent areas cause problems with their method, but it's not clear why ghost imaging fails for this kind of object. Or is the failure restricted to the authors' specific implementation of ghost imaging?
7) Lines 191-203 is a duplicate paragraph of the previous one
Author Response
Dear reviewer 3,
I appreciate the reviewer 3 for many helpful and appropriate comments.
We carefully revised our manuscript in accordance with your comments.
I send the attached document for our reply.
Yours sincerely,
Author Response File: Author Response.docx
Round 2
Reviewer 1 Report
The authors have resubmitted a revised version of their manuscript on computational ghost imaging using VCSEL arrays. The manuscript has improved somewhat, although the responses to the referee comments are lacking. For example the authors' remark that a ground glass disk would not work for CGI, while a designed DOE would work, is obviously off. In their work the authors actively measure the phase response anyway, so it would not be a problem to measure the phase change introduced by a ground glass disk and use it as a DOE. Also in several other respects the manuscript is not as clear as it could be with respect to the language used and a number of typos that is still present. However, the scientific content is valid and how clear they want the manuscript to be is the choice of the authors.
Accordingly, I support publication of the manucript, but suggest that the authors again have a look at typos and the ease of reading the text.
Author Response
Dear reviewer 1,
I appreciate your review very much, again.
Our response for your comments is in the attached file.
Yours sincerely,
Author Response File: Author Response.docx
Reviewer 2 Report
Dear Editor,
the authors have revised the apper according my requests and it is now suited to be published, after a couple of minus comments are considered:
the sentence "A system which is satisfied by the above conditions is considered to be able to be constructed. " should be revised ref 12 mentioned in the paper does not correspond to what reported in the list of referencesAuthor Response
Dear reviewer 2,
I appreciate your review very much, again.
Our response for your comments is in the attached file.
Yours sincerely,
Author Response File: Author Response.docx
